diff --git a/src/.vuepress/sidebar/V1.3.x/en.ts b/src/.vuepress/sidebar/V1.3.x/en.ts index c22dd039b..1d2d30feb 100644 --- a/src/.vuepress/sidebar/V1.3.x/en.ts +++ b/src/.vuepress/sidebar/V1.3.x/en.ts @@ -104,34 +104,24 @@ export const enSidebar = { text: 'Advanced Features', collapsible: true, prefix: 'User-Manual/', - // children: 'structure', children: [ + { text: 'Data Sync(V1.3.0/1/2)', link: 'Data-Sync-old_apache' }, + { text: 'Data Sync(V1.3.3)', link: 'Data-Sync_apache' }, + { text: 'Data Subscription', link: 'Data-subscription' }, { - text: 'Data Sync', - collapsible: true, - children: [ - { text: 'Data Sync(V1.3.0/1/2)', link: 'Data-Sync-old_apache' }, - { text: 'Data Sync(V1.3.3)', link: 'Data-Sync_apache' }, - ], - }, - { text: 'Data Subscription(V1.3.4)', link: 'Data-subscription' }, - { text: 'AI Capability', link: 'AINode_apache' }, - { - text: 'Security Management', + text: 'Stream Computing', collapsible: true, children: [ - { text: 'Authority Management', link: 'Authority-Management' }, + { text: 'Stream Computing Framework', link: 'Streaming_apache' }, + { text: 'Continuous Query', link: 'Database-Programming' }, + { text: 'Trigger', link: 'Trigger' }, ], }, { text: 'UDF', link: 'User-defined-function_apache' }, - { text: 'Continuous Query', link: 'Database-Programming' }, { - text: 'Database Programming', + text: 'Security Permissions', collapsible: true, - children: [ - { text: 'Trigger', link: 'Trigger' }, - { text: 'Stream Processing', link: 'Streaming_apache' }, - ], + children: [{ text: 'Permission Management', link: 'Authority-Management' }], }, { text: 'Maintenance SQL', @@ -144,6 +134,15 @@ export const enSidebar = { }, ], }, + { + text: 'AI capability', + collapsible: true, + prefix: 'AI-capability/', + children: [ + { text: 'AINode', link: 'AINode_apache' }, + { text: 'TimeSeries Large Model', link: 'TimeSeries-Large-Model' }, + ], + }, { text: 'Tools System', collapsible: true, diff --git a/src/.vuepress/sidebar/V1.3.x/zh.ts b/src/.vuepress/sidebar/V1.3.x/zh.ts index f90d9abfe..b4473ead4 100644 --- a/src/.vuepress/sidebar/V1.3.x/zh.ts +++ b/src/.vuepress/sidebar/V1.3.x/zh.ts @@ -93,33 +93,25 @@ export const zhSidebar = { text: '高级功能', collapsible: true, prefix: 'User-Manual/', - // children: 'structure', children: [ + { text: '数据同步(V1.3.0/1/2)', link: 'Data-Sync-old_apache' }, + { text: '数据同步(V1.3.3)', link: 'Data-Sync_apache' }, + { text: '数据订阅', link: 'Data-subscription' }, { - text: '数据同步', + text: '流计算', collapsible: true, children: [ - { text: '数据同步(V1.3.0/1/2)', link: 'Data-Sync-old_apache' }, - { text: '数据同步(V1.3.3)', link: 'Data-Sync_apache' }, + { text: '流计算框架', link: 'Streaming_apache' }, + { text: '连续查询', link: 'Database-Programming' }, + { text: '触发器', link: 'Trigger' }, ], }, - { text: '数据订阅(V1.3.4)', link: 'Data-subscription' }, - { text: 'AI能力', link: 'AINode_apache' }, + { text: 'UDF', link: 'User-defined-function_apache' }, { - text: '安全管理', + text: '安全权限', collapsible: true, children: [{ text: '权限管理', link: 'Authority-Management' }], }, - { text: '用户自定义函数', link: 'User-defined-function_apache' }, - { text: '连续查询', link: 'Database-Programming' }, - { - text: '数据库编程', - collapsible: true, - children: [ - { text: '触发器', link: 'Trigger' }, - { text: '流处理框架', link: 'Streaming_apache' }, - ], - }, { text: '运维语句', collapsible: true, @@ -131,6 +123,15 @@ export const zhSidebar = { }, ], }, + { + text: 'AI 能力', + collapsible: true, + prefix: 'AI-capability/', + children: [ + { text: 'AINode', link: 'AINode_apache' }, + { text: '时序大模型', link: 'TimeSeries-Large-Model' }, + ], + }, { text: '工具体系', collapsible: true, diff --git a/src/.vuepress/sidebar/V2.0.x/en-Tree.ts b/src/.vuepress/sidebar/V2.0.x/en-Tree.ts index cdc6f3610..4efcd8b9c 100644 --- a/src/.vuepress/sidebar/V2.0.x/en-Tree.ts +++ b/src/.vuepress/sidebar/V2.0.x/en-Tree.ts @@ -100,21 +100,19 @@ export const enSidebar = { { text: 'Data Sync', link: 'Data-Sync_apache' }, { text: 'Data Subscription', link: 'Data-subscription' }, { - text: 'Security Management', + text: 'Stream Computing', collapsible: true, children: [ - { text: 'Authority Management', link: 'Authority-Management' }, + { text: 'Stream Computing Framework', link: 'Streaming_apache' }, + { text: 'Continuous Query', link: 'Database-Programming' }, + { text: 'Trigger', link: 'Trigger' }, ], }, { text: 'UDF', link: 'User-defined-function_apache' }, - { text: 'Continuous Query', link: 'Database-Programming' }, { - text: 'Database Programming', + text: 'Security Permissions', collapsible: true, - children: [ - { text: 'Trigger', link: 'Trigger' }, - { text: 'Stream Processing', link: 'Streaming_apache' }, - ], + children: [{ text: 'Permission Management', link: 'Authority-Management' }], }, { text: 'Maintenance SQL', diff --git a/src/.vuepress/sidebar/V2.0.x/zh-Tree.ts b/src/.vuepress/sidebar/V2.0.x/zh-Tree.ts index 1b695b103..bc357d987 100644 --- a/src/.vuepress/sidebar/V2.0.x/zh-Tree.ts +++ b/src/.vuepress/sidebar/V2.0.x/zh-Tree.ts @@ -89,20 +89,20 @@ export const zhSidebar = { { text: '数据同步', link: 'Data-Sync_apache' }, { text: '数据订阅', link: 'Data-subscription' }, { - text: '安全管理', - collapsible: true, - children: [{ text: '权限管理', link: 'Authority-Management' }], - }, - { text: '用户自定义函数', link: 'User-defined-function_apache' }, - { text: '连续查询', link: 'Database-Programming' }, - { - text: '数据库编程', + text: '流计算', collapsible: true, children: [ + { text: '流计算框架', link: 'Streaming_apache' }, + { text: '连续查询', link: 'Database-Programming' }, { text: '触发器', link: 'Trigger' }, - { text: '流处理框架', link: 'Streaming_apache' }, ], }, + { text: 'UDF', link: 'User-defined-function_apache' }, + { + text: '安全权限', + collapsible: true, + children: [{ text: '权限管理', link: 'Authority-Management' }], + }, { text: '运维语句', collapsible: true, diff --git a/src/.vuepress/sidebar_timecho/V1.3.x/en.ts b/src/.vuepress/sidebar_timecho/V1.3.x/en.ts index ff1ea8475..43ddb7535 100644 --- a/src/.vuepress/sidebar_timecho/V1.3.x/en.ts +++ b/src/.vuepress/sidebar_timecho/V1.3.x/en.ts @@ -113,37 +113,29 @@ export const enSidebar = { text: 'Advanced Features', collapsible: true, prefix: 'User-Manual/', - // children: 'structure', children: [ + { text: 'Data Sync(V1.3.0/1/2)', link: 'Data-Sync-old_timecho' }, + { text: 'Data Sync(V1.3.3)', link: 'Data-Sync_timecho' }, + { text: 'Data Subscription', link: 'Data-subscription' }, { - text: 'Data Sync', + text: 'Stream Computing', collapsible: true, children: [ - { text: 'Data Sync(V1.3.0/1/2)', link: 'Data-Sync-old_timecho' }, - { text: 'Data Sync(V1.3.3)', link: 'Data-Sync_timecho' }, - ], - }, - { text: 'Data Subscription(V1.3.4)', link: 'Data-subscription' }, - { text: 'AI Capability', link: 'AINode_timecho' }, - { - text: 'Security Management', - collapsible: true, - children: [ - { text: 'White List', link: 'White-List_timecho' }, - { text: 'Audit Log', link: 'Audit-Log_timecho' }, - { text: 'Authority Management', link: 'Authority-Management' }, + { text: 'Stream Computing Framework', link: 'Streaming_timecho' }, + { text: 'Continuous Query', link: 'Database-Programming' }, + { text: 'Trigger', link: 'Trigger' }, ], }, + { text: 'Tiered Storage', link: 'Tiered-Storage_timecho' }, { text: 'UDF', link: 'User-defined-function_timecho' }, { text: 'View', link: 'IoTDB-View_timecho' }, - { text: 'Tiered Storage', link: 'Tiered-Storage_timecho' }, - { text: 'Continuous Query', link: 'Database-Programming' }, { - text: 'Database Programming', + text: 'Security Permissions', collapsible: true, children: [ - { text: 'Trigger', link: 'Trigger' }, - { text: 'Stream Processing', link: 'Streaming_timecho' }, + { text: 'Permission Management', link: 'Authority-Management' }, + { text: 'White List', link: 'White-List_timecho' }, + { text: 'Security Audit', link: 'Audit-Log_timecho' }, ], }, { @@ -157,6 +149,15 @@ export const enSidebar = { }, ], }, + { + text: 'AI capability', + collapsible: true, + prefix: 'AI-capability/', + children: [ + { text: 'AINode', link: 'AINode_timecho' }, + { text: 'TimeSeries Large Model', link: 'TimeSeries-Large-Model' }, + ], + }, { text: 'Tools System', collapsible: true, diff --git a/src/.vuepress/sidebar_timecho/V1.3.x/zh.ts b/src/.vuepress/sidebar_timecho/V1.3.x/zh.ts index 83f53f0ca..0f70c2853 100644 --- a/src/.vuepress/sidebar_timecho/V1.3.x/zh.ts +++ b/src/.vuepress/sidebar_timecho/V1.3.x/zh.ts @@ -96,37 +96,29 @@ export const zhSidebar = { text: '高级功能', collapsible: true, prefix: 'User-Manual/', - // children: 'structure', children: [ + { text: '数据同步(V1.3.0/1/2)', link: 'Data-Sync-old_timecho' }, + { text: '数据同步(V1.3.3)', link: 'Data-Sync_timecho' }, + { text: '数据订阅', link: 'Data-subscription' }, { - text: '数据同步', - collapsible: true, - children: [ - { text: '数据同步(V1.3.0/1/2)', link: 'Data-Sync-old_timecho' }, - { text: '数据同步(V1.3.3)', link: 'Data-Sync_timecho' }, - ], - }, - { text: '数据订阅(V1.3.4)', link: 'Data-subscription' }, - { text: 'AI能力', link: 'AINode_timecho' }, - { - text: '安全管理', + text: '流计算', collapsible: true, children: [ - { text: '白名单', link: 'White-List_timecho' }, - { text: '审计日志', link: 'Audit-Log_timecho' }, - { text: '权限管理', link: 'Authority-Management' }, + { text: '流计算框架', link: 'Streaming_timecho' }, + { text: '连续查询', link: 'Database-Programming' }, + { text: '触发器', link: 'Trigger' }, ], }, - { text: '用户自定义函数', link: 'User-defined-function_timecho' }, - { text: '视图', link: 'IoTDB-View_timecho' }, { text: '多级存储', link: 'Tiered-Storage_timecho' }, - { text: '连续查询', link: 'Database-Programming' }, + { text: 'UDF', link: 'User-defined-function_timecho' }, + { text: '视图', link: 'IoTDB-View_timecho' }, { - text: '数据库编程', + text: '安全权限', collapsible: true, children: [ - { text: '触发器', link: 'Trigger' }, - { text: '流处理框架', link: 'Streaming_timecho' }, + { text: '权限管理', link: 'Authority-Management' }, + { text: '白名单', link: 'White-List_timecho' }, + { text: '安全审计', link: 'Audit-Log_timecho' }, ], }, { @@ -140,6 +132,15 @@ export const zhSidebar = { }, ], }, + { + text: 'AI 能力', + collapsible: true, + prefix: 'AI-capability/', + children: [ + { text: 'AINode', link: 'AINode_timecho' }, + { text: '时序大模型', link: 'TimeSeries-Large-Model' }, + ], + }, { text: '工具体系', collapsible: true, diff --git a/src/.vuepress/sidebar_timecho/V2.0.x/en-Tree.ts b/src/.vuepress/sidebar_timecho/V2.0.x/en-Tree.ts index 5e7e1f80e..b64bebe71 100644 --- a/src/.vuepress/sidebar_timecho/V2.0.x/en-Tree.ts +++ b/src/.vuepress/sidebar_timecho/V2.0.x/en-Tree.ts @@ -109,24 +109,24 @@ export const enSidebar = { { text: 'Data Sync', link: 'Data-Sync_timecho' }, { text: 'Data Subscription', link: 'Data-subscription' }, { - text: 'Security Management', + text: 'Stream Computing', collapsible: true, children: [ - { text: 'White List', link: 'White-List_timecho' }, - { text: 'Audit Log', link: 'Audit-Log_timecho' }, - { text: 'Authority Management', link: 'Authority-Management' }, + { text: 'Stream Computing Framework', link: 'Streaming_timecho' }, + { text: 'Continuous Query', link: 'Database-Programming' }, + { text: 'Trigger', link: 'Trigger' }, ], }, + { text: 'Tiered Storage', link: 'Tiered-Storage_timecho' }, { text: 'UDF', link: 'User-defined-function_timecho' }, { text: 'View', link: 'IoTDB-View_timecho' }, - { text: 'Tiered Storage', link: 'Tiered-Storage_timecho' }, - { text: 'Continuous Query', link: 'Database-Programming' }, { - text: 'Database Programming', + text: 'Security Permissions', collapsible: true, children: [ - { text: 'Trigger', link: 'Trigger' }, - { text: 'Stream Processing', link: 'Streaming_timecho' }, + { text: 'Permission Management', link: 'Authority-Management' }, + { text: 'White List', link: 'White-List_timecho' }, + { text: 'Security Audit', link: 'Audit-Log_timecho' }, ], }, { diff --git a/src/.vuepress/sidebar_timecho/V2.0.x/zh-Tree.ts b/src/.vuepress/sidebar_timecho/V2.0.x/zh-Tree.ts index 1dfd77724..2a686ff66 100644 --- a/src/.vuepress/sidebar_timecho/V2.0.x/zh-Tree.ts +++ b/src/.vuepress/sidebar_timecho/V2.0.x/zh-Tree.ts @@ -92,24 +92,24 @@ export const zhSidebar = { { text: '数据同步', link: 'Data-Sync_timecho' }, { text: '数据订阅', link: 'Data-subscription' }, { - text: '安全管理', + text: '流计算', collapsible: true, children: [ - { text: '白名单', link: 'White-List_timecho' }, - { text: '审计日志', link: 'Audit-Log_timecho' }, - { text: '权限管理', link: 'Authority-Management' }, + { text: '流计算框架', link: 'Streaming_timecho' }, + { text: '连续查询', link: 'Database-Programming' }, + { text: '触发器', link: 'Trigger' }, ], }, - { text: '用户自定义函数', link: 'User-defined-function_timecho' }, - { text: '视图', link: 'IoTDB-View_timecho' }, { text: '多级存储', link: 'Tiered-Storage_timecho' }, - { text: '连续查询', link: 'Database-Programming' }, + { text: 'UDF', link: 'User-defined-function_timecho' }, + { text: '视图', link: 'IoTDB-View_timecho' }, { - text: '数据库编程', + text: '安全权限', collapsible: true, children: [ - { text: '触发器', link: 'Trigger' }, - { text: '流处理框架', link: 'Streaming_timecho' }, + { text: '权限管理', link: 'Authority-Management' }, + { text: '白名单', link: 'White-List_timecho' }, + { text: '安全审计', link: 'Audit-Log_timecho' }, ], }, { diff --git a/src/UserGuide/Master/Tree/User-Manual/Audit-Log_timecho.md b/src/UserGuide/Master/Tree/User-Manual/Audit-Log_timecho.md index 77faf0a7a..61ba509c5 100644 --- a/src/UserGuide/Master/Tree/User-Manual/Audit-Log_timecho.md +++ b/src/UserGuide/Master/Tree/User-Manual/Audit-Log_timecho.md @@ -19,7 +19,7 @@ --> -# Audit log +# Security Audit ## 1. Background of the function diff --git a/src/UserGuide/Master/Tree/User-Manual/Streaming_apache.md b/src/UserGuide/Master/Tree/User-Manual/Streaming_apache.md index da81e199f..b5a8e2419 100644 --- a/src/UserGuide/Master/Tree/User-Manual/Streaming_apache.md +++ b/src/UserGuide/Master/Tree/User-Manual/Streaming_apache.md @@ -19,7 +19,7 @@ --> -# Stream Processing +# Stream Computing Framework The IoTDB stream processing framework allows users to implement customized stream processing logic, which can monitor and capture storage engine changes, transform changed data, and push transformed data outward. diff --git a/src/UserGuide/Master/Tree/User-Manual/Streaming_timecho.md b/src/UserGuide/Master/Tree/User-Manual/Streaming_timecho.md index 3f72c1f8f..07a3e3017 100644 --- a/src/UserGuide/Master/Tree/User-Manual/Streaming_timecho.md +++ b/src/UserGuide/Master/Tree/User-Manual/Streaming_timecho.md @@ -19,7 +19,7 @@ --> -# Stream Processing +# Stream Computing Framework The IoTDB stream processing framework allows users to implement customized stream processing logic, which can monitor and capture storage engine changes, transform changed data, and push transformed data outward. diff --git a/src/UserGuide/V1.2.x/User-Manual/Security-Management_timecho.md b/src/UserGuide/V1.2.x/User-Manual/Security-Management_timecho.md index baf577524..4e88a0183 100644 --- a/src/UserGuide/V1.2.x/User-Manual/Security-Management_timecho.md +++ b/src/UserGuide/V1.2.x/User-Manual/Security-Management_timecho.md @@ -25,7 +25,6 @@ coming soon -## Audit Log - +## Security Audit coming soon diff --git a/src/UserGuide/V1.2.x/User-Manual/Streaming.md b/src/UserGuide/V1.2.x/User-Manual/Streaming.md index ff21b4e71..4ac8ec837 100644 --- a/src/UserGuide/V1.2.x/User-Manual/Streaming.md +++ b/src/UserGuide/V1.2.x/User-Manual/Streaming.md @@ -19,7 +19,7 @@ --> -# Stream Processing +# Stream Computing Framework The IoTDB stream processing framework allows users to implement customized stream processing logic, which can monitor and capture storage engine changes, transform changed data, and push transformed data outward. diff --git a/src/UserGuide/V1.2.x/User-Manual/Streaming_timecho.md b/src/UserGuide/V1.2.x/User-Manual/Streaming_timecho.md index 99b6d243d..9dda4c279 100644 --- a/src/UserGuide/V1.2.x/User-Manual/Streaming_timecho.md +++ b/src/UserGuide/V1.2.x/User-Manual/Streaming_timecho.md @@ -19,7 +19,7 @@ --> -# Stream Processing +# Stream Computing Framework The IoTDB stream processing framework allows users to implement customized stream processing logic, which can monitor and capture storage engine changes, transform changed data, and push transformed data outward. diff --git a/src/UserGuide/V1.3.x/User-Manual/AINode_apache.md b/src/UserGuide/V1.3.x/AI-capability/AINode_apache.md similarity index 98% rename from src/UserGuide/V1.3.x/User-Manual/AINode_apache.md rename to src/UserGuide/V1.3.x/AI-capability/AINode_apache.md index 3045fa625..95fcccd35 100644 --- a/src/UserGuide/V1.3.x/User-Manual/AINode_apache.md +++ b/src/UserGuide/V1.3.x/AI-capability/AINode_apache.md @@ -19,7 +19,7 @@ --> -# AI Capability（AINode） +# AINode AINode is the third internal node after ConfigNode and DataNode in Apache IoTDB, which extends the capability of machine learning analysis of time series by interacting with DataNode and ConfigNode of IoTDB cluster, supports the introduction of pre-existing machine learning models from the outside to be registered, and uses the registered models in the It supports the process of introducing existing machine learning models from outside for registration, and using the registered models to complete the time series analysis tasks on the specified time series data through simple SQL statements, which integrates the model creation, management and inference in the database engine. At present, we have provided machine learning algorithms or self-developed models for common timing analysis scenarios (e.g. prediction and anomaly detection). @@ -33,7 +33,7 @@ The responsibilities of the three nodes are as follows: - **DataNode**: responsible for receiving and parsing SQL requests from users; responsible for storing time-series data; responsible for preprocessing computation of data. - **AINode**: responsible for model file import creation and model inference. -## Advantageous features +## 1. Advantageous features Compared with building a machine learning service alone, it has the following advantages: @@ -50,7 +50,7 @@ Compared with building a machine learning service alone, it has the following ad -## Basic Concepts +## 2. Basic Concepts - **Model**: a machine learning model that takes time-series data as input and outputs the results or decisions of an analysis task. Model is the basic management unit of AINode, which supports adding (registration), deleting, checking, and using (inference) of models. - **Create**: Load externally designed or trained model files or algorithms into MLNode for unified management and use by IoTDB. @@ -61,16 +61,16 @@ Compared with building a machine learning service alone, it has the following ad :::: -## Installation and Deployment +## 3. Installation and Deployment The deployment of AINode can be found in the document [Deployment Guidelines](../Deployment-and-Maintenance/AINode_Deployment_apache.md#ainode-deployment) . -## Usage Guidelines +## 4. Usage Guidelines AINode provides model creation and deletion process for deep learning models related to timing data. Built-in models do not need to be created and deleted, they can be used directly, and the built-in model instances created after inference is completed will be destroyed automatically. -### Registering Models +### 4.1 Registering Models A trained deep learning model can be registered by specifying the vector dimensions of the model's inputs and outputs, which can be used for model inference. @@ -156,7 +156,7 @@ After the SQL is executed, the registration process will be carried out asynchro Once the model registration is complete, you can call specific functions and perform model inference by using normal queries. -### Viewing Models +### 4.2 Viewing Models Successfully registered models can be queried for model-specific information through the show models command. The SQL definition is as follows: @@ -204,7 +204,7 @@ IoTDB> show models We have registered the corresponding model earlier, you can view the model status through the corresponding designation, active indicates that the model is successfully registered and can be used for inference. -### Delete Model +### 4.3 Delete Model For a successfully registered model, the user can delete it via SQL. In addition to deleting the meta information on the configNode, this operation also deletes all the related model files under the AINode. The SQL is as follows: @@ -214,7 +214,7 @@ drop model You need to specify the model model_name that has been successfully registered to delete the corresponding model. Since model deletion involves the deletion of data on multiple nodes, the operation will not be completed immediately, and the state of the model at this time is DROPPING, and the model in this state cannot be used for model inference. -### Using Built-in Model Reasoning +### 4.4 Using Built-in Model Reasoning The SQL syntax is as follows: @@ -284,7 +284,7 @@ IoTDB> call inference(_Stray, "select s0 from root.eg.airline", k=2) Total line number = 144 ``` -### Reasoning with Deep Learning Models +### 4.5 Reasoning with Deep Learning Models The SQL syntax is as follows: @@ -444,7 +444,7 @@ Total line number = 4 In the result set, each row's label corresponds to the output of the anomaly detection model after inputting each group of 24 rows of data. -## Privilege Management +## 5. Privilege Management When using AINode related functions, the authentication of IoTDB itself can be used to do a permission management, users can only use the model management related functions when they have the USE_MODEL permission. When using the inference function, the user needs to have the permission to access the source sequence corresponding to the SQL of the input model. @@ -453,9 +453,9 @@ When using AINode related functions, the authentication of IoTDB itself can be u | USE_MODEL | create model/show models/drop model | √ | √ | x | | READ_DATA| call inference | √ | √|√ | -## Practical Examples +## 6. Practical Examples -### Power Load Prediction +### 6.1 Power Load Prediction In some industrial scenarios, there is a need to predict power loads, which can be used to optimise power supply, conserve energy and resources, support planning and expansion, and enhance power system reliability. @@ -526,7 +526,7 @@ The data before 10/24 00:00 represents the past data input to the model, the blu As can be seen, we have used the relationship between the six load information and the corresponding time oil temperatures for the past 96 hours (4 days) to model the possible changes in this data for the oil temperature for the next 48 hours (2 days) based on the inter-relationships between the sequences learned previously, and it can be seen that the predicted curves maintain a high degree of consistency in trend with the actual results after visualisation. -### Power Prediction +### 6.2 Power Prediction Power monitoring of current, voltage and power data is required in substations for detecting potential grid problems, identifying faults in the power system, effectively managing grid loads and analysing power system performance and trends. @@ -592,7 +592,7 @@ The data before 02/14 20:48 represents the past data input to the model, the blu It can be seen that we used the voltage data from the past 10 minutes and, based on the previously learned inter-sequence relationships, modeled the possible changes in the phase C voltage data for the next 5 minutes. The visualized forecast curve shows a certain degree of synchronicity with the actual results in terms of trend. -### Anomaly Detection +### 6.3 Anomaly Detection In the civil aviation and transport industry, there exists a need for anomaly detection of the number of passengers travelling on an aircraft. The results of anomaly detection can be used to guide the adjustment of flight scheduling to make the organisation more efficient. diff --git a/src/UserGuide/V1.3.x/User-Manual/AINode_timecho.md b/src/UserGuide/V1.3.x/AI-capability/AINode_timecho.md similarity index 98% rename from src/UserGuide/V1.3.x/User-Manual/AINode_timecho.md rename to src/UserGuide/V1.3.x/AI-capability/AINode_timecho.md index b797c8dfb..204f91cf9 100644 --- a/src/UserGuide/V1.3.x/User-Manual/AINode_timecho.md +++ b/src/UserGuide/V1.3.x/AI-capability/AINode_timecho.md @@ -19,7 +19,7 @@ --> -# AI Capability（AINode） +# AINode AINode is the third internal node after ConfigNode and DataNode in Apache IoTDB, which extends the capability of machine learning analysis of time series by interacting with DataNode and ConfigNode of IoTDB cluster, supports the introduction of pre-existing machine learning models from the outside to be registered, and uses the registered models in the It supports the process of introducing existing machine learning models from outside for registration, and using the registered models to complete the time series analysis tasks on the specified time series data through simple SQL statements, which integrates the model creation, management and inference in the database engine. At present, we have provided machine learning algorithms or self-developed models for common timing analysis scenarios (e.g. prediction and anomaly detection). @@ -33,7 +33,7 @@ The responsibilities of the three nodes are as follows: - **DataNode**: responsible for receiving and parsing SQL requests from users; responsible for storing time-series data; responsible for preprocessing computation of data. - **AINode**: responsible for model file import creation and model inference. -## Advantageous features +## 1. Advantageous features Compared with building a machine learning service alone, it has the following advantages: @@ -50,7 +50,7 @@ Compared with building a machine learning service alone, it has the following ad -## Basic Concepts +## 2. Basic Concepts - **Model**: a machine learning model that takes time-series data as input and outputs the results or decisions of an analysis task. Model is the basic management unit of AINode, which supports adding (registration), deleting, checking, and using (inference) of models. - **Create**: Load externally designed or trained model files or algorithms into MLNode for unified management and use by IoTDB. @@ -61,16 +61,16 @@ Compared with building a machine learning service alone, it has the following ad :::: -## Installation and Deployment +## 3. Installation and Deployment The deployment of AINode can be found in the document [Deployment Guidelines](../Deployment-and-Maintenance/AINode_Deployment_timecho.md#AINode-部署) . -## Usage Guidelines +## 4. Usage Guidelines AINode provides model creation and deletion process for deep learning models related to timing data. Built-in models do not need to be created and deleted, they can be used directly, and the built-in model instances created after inference is completed will be destroyed automatically. -### Registering Models +### 4.1 Registering Models A trained deep learning model can be registered by specifying the vector dimensions of the model's inputs and outputs, which can be used for model inference. @@ -156,7 +156,7 @@ After the SQL is executed, the registration process will be carried out asynchro Once the model registration is complete, you can call specific functions and perform model inference by using normal queries. -### Viewing Models +### 4.2 Viewing Models Successfully registered models can be queried for model-specific information through the show models command. The SQL definition is as follows: @@ -204,7 +204,7 @@ IoTDB> show models We have registered the corresponding model earlier, you can view the model status through the corresponding designation, active indicates that the model is successfully registered and can be used for inference. -### Delete Model +### 4.3 Delete Model For a successfully registered model, the user can delete it via SQL. In addition to deleting the meta information on the configNode, this operation also deletes all the related model files under the AINode. The SQL is as follows: @@ -214,7 +214,7 @@ drop model You need to specify the model model_name that has been successfully registered to delete the corresponding model. Since model deletion involves the deletion of data on multiple nodes, the operation will not be completed immediately, and the state of the model at this time is DROPPING, and the model in this state cannot be used for model inference. -### Using Built-in Model Reasoning +### 4.4 Using Built-in Model Reasoning The SQL syntax is as follows: @@ -284,7 +284,7 @@ IoTDB> call inference(_Stray, "select s0 from root.eg.airline", k=2) Total line number = 144 ``` -### Reasoning with Deep Learning Models +### 4.5 Reasoning with Deep Learning Models The SQL syntax is as follows: @@ -444,7 +444,13 @@ Total line number = 4 In the result set, each row's label corresponds to the output of the anomaly detection model after inputting each group of 24 rows of data. -## Privilege Management + +### 4.6 TimeSeries Large Models Import Steps + +AINode currently supports a variety of time series large models. For deployment and usage, please refer to [TimeSeries Large Models](../AI-capability/TimeSeries-Large-Model) + + +## 5. Privilege Management When using AINode related functions, the authentication of IoTDB itself can be used to do a permission management, users can only use the model management related functions when they have the USE_MODEL permission. When using the inference function, the user needs to have the permission to access the source sequence corresponding to the SQL of the input model. @@ -453,9 +459,9 @@ When using AINode related functions, the authentication of IoTDB itself can be u | USE_MODEL | create model/show models/drop model | √ | √ | x | | READ_DATA| call inference | √ | √|√ | -## Practical Examples +## 6. Practical Examples -### Power Load Prediction +### 6.1 Power Load Prediction In some industrial scenarios, there is a need to predict power loads, which can be used to optimise power supply, conserve energy and resources, support planning and expansion, and enhance power system reliability. @@ -526,7 +532,7 @@ The data before 10/24 00:00 represents the past data input to the model, the blu As can be seen, we have used the relationship between the six load information and the corresponding time oil temperatures for the past 96 hours (4 days) to model the possible changes in this data for the oil temperature for the next 48 hours (2 days) based on the inter-relationships between the sequences learned previously, and it can be seen that the predicted curves maintain a high degree of consistency in trend with the actual results after visualisation. -### Power Prediction +### 6.2 Power Prediction Power monitoring of current, voltage and power data is required in substations for detecting potential grid problems, identifying faults in the power system, effectively managing grid loads and analysing power system performance and trends. @@ -592,7 +598,7 @@ The data before 02/14 20:48 represents the past data input to the model, the blu It can be seen that we used the voltage data from the past 10 minutes and, based on the previously learned inter-sequence relationships, modeled the possible changes in the phase C voltage data for the next 5 minutes. The visualized forecast curve shows a certain degree of synchronicity with the actual results in terms of trend. -### Anomaly Detection +### 6.3 Anomaly Detection In the civil aviation and transport industry, there exists a need for anomaly detection of the number of passengers travelling on an aircraft. The results of anomaly detection can be used to guide the adjustment of flight scheduling to make the organisation more efficient. diff --git a/src/UserGuide/V1.3.x/AI-capability/TimeSeries-Large-Model.md b/src/UserGuide/V1.3.x/AI-capability/TimeSeries-Large-Model.md new file mode 100644 index 000000000..4ce2801d0 --- /dev/null +++ b/src/UserGuide/V1.3.x/AI-capability/TimeSeries-Large-Model.md @@ -0,0 +1,117 @@ + + +# TimeSeries Large Model + +## Introduction + +A time series large model is a foundational model specifically designed for time series analysis. The IoTDB team has independently developed time series large models, which are pre-trained on massive time series data using technologies such as transformer structures. These models can understand and generate time series data across various domains and are applicable to applications like time series forecasting, anomaly detection, and time series imputation. Unlike traditional time series analysis techniques, time series large models possess the capability to extract universal features and provide technical services based on zero-shot analysis and fine-tuning for a wide range of analytical tasks. + +The team's related technologies of time series large models have been published in top international machine learning conferences. + +## Application Scenarios + +- **Time Series Forecasting**: Provides forecasting services for time series data in industrial production, natural environments, and other areas, helping users to understand future trends in advance. +- **Data Imputation**: For missing segments in time series, perform context imputation to enhance the continuity and completeness of the dataset. +- **Anomaly Detection**: Utilizing regression analysis technology, monitor time series data in real-time and provide timely warnings for potential anomalies. + +![](/img/LargeModel10.png) + +## Timer Model + +The Timer model not only demonstrates excellent few-shot generalization and multi-task adaptation capabilities but also gains a rich knowledge base through pre-training, endowing it with the universal capability to handle a variety of downstream tasks, featuring the following: + +- **Generalization**: The model can be fine-tuned using a small number of samples to achieve leading predictive performance in the industry. +- **Versatility**: The model is designed flexibly to adapt to various task requirements and supports variable input and output lengths, enabling it to play a role in various application scenarios. +- **Scalability**: As the number of model parameters increases or the scale of pre-training data expands, the model's performance continues to improve, ensuring the model can optimize its predictive effects with the growth of time and data volume. + +![](/img/LargeModel02.png) + +## Timer-XL Model + +Timer-XL is an upgraded version of Timer that further extends the network structure and achieves comprehensive breakthroughs in multiple dimensions: + +- **Long Context Support**: This model breaks through the limitations of traditional time series forecasting models, supporting the processing of thousands of tokens (equivalent to tens of thousands of time points) of input, effectively addressing the bottleneck of context length. +- **Multi-variable Forecasting Scenario Coverage**: Supports a variety of forecasting scenarios, including non-stationary time series forecasting, multi-variable prediction tasks, and predictions involving covariates, meeting diverse business needs. +- **Large-scale Industrial Time Series Dataset**: Pre-trained using a massive industrial IoT time series dataset that has a large volume, excellent quality, and rich domain characteristics, covering energy, aerospace, steel, transportation, and more. + + +## Effect Demonstration + +Time series large models can adapt to real time series data from various fields and scenarios, showing excellent processing effects in various tasks. Here are the real performances on different data: + +**Time Series Forecasting:** + +Utilizing the predictive capabilities of the time series large model, it can accurately predict the future trend of time series. As shown in the figure, the blue curve represents the predicted trend, and the red curve represents the actual trend, with the two curves highly matching. + +![](/img/LargeModel03.png) + +**Data Imputation:**： + +Using the time series large model to perform predictive imputation for missing data segments. + +![](/img/LargeModel04.png) + + +**Anomaly Detection:**： + +Utilizing the time series large model to accurately identify anomalies that deviate significantly from the normal trend. + +![](/img/LargeModel05.png) + +## Deployment Usage + +1. Open the IoTDB CLI console and verify that the ConfigNode, DataNode, and AINode statuses are all ​Running. + +Check command: + +```sql +show cluster +``` + +![](/img/ainode-timer-1.png) + +2. Model file storage path: It is recommended to place the model files in the same directory as the AINode installation package. + You may create a new folder to store model files. + +3. Register the model + +Use the following SQL statement: + +```sql +create model using uri +``` + +Example (for the Timer model): + +```sql +create model Timer using uri +``` + +4. Verify model registration success + +Check command: + +```sql +show models +``` + +![](/img/LargeModel06.png) diff --git a/src/UserGuide/V1.3.x/User-Manual/Audit-Log_timecho.md b/src/UserGuide/V1.3.x/User-Manual/Audit-Log_timecho.md index d3ff3aae6..976b7da17 100644 --- a/src/UserGuide/V1.3.x/User-Manual/Audit-Log_timecho.md +++ b/src/UserGuide/V1.3.x/User-Manual/Audit-Log_timecho.md @@ -19,7 +19,7 @@ --> -# Audit log +# Security Audit ## Background of the function diff --git a/src/UserGuide/V1.3.x/User-Manual/Streaming_apache.md b/src/UserGuide/V1.3.x/User-Manual/Streaming_apache.md index cfb6a3433..7f9b9716a 100644 --- a/src/UserGuide/V1.3.x/User-Manual/Streaming_apache.md +++ b/src/UserGuide/V1.3.x/User-Manual/Streaming_apache.md @@ -19,7 +19,7 @@ --> -# Stream Processing +# Stream Computing Framework The IoTDB stream processing framework allows users to implement customized stream processing logic, which can monitor and capture storage engine changes, transform changed data, and push transformed data outward. diff --git a/src/UserGuide/V1.3.x/User-Manual/Streaming_timecho.md b/src/UserGuide/V1.3.x/User-Manual/Streaming_timecho.md index 8d6f50e44..80edebe9c 100644 --- a/src/UserGuide/V1.3.x/User-Manual/Streaming_timecho.md +++ b/src/UserGuide/V1.3.x/User-Manual/Streaming_timecho.md @@ -19,7 +19,7 @@ --> -# Stream Processing +# Stream Computing Framework The IoTDB stream processing framework allows users to implement customized stream processing logic, which can monitor and capture storage engine changes, transform changed data, and push transformed data outward. diff --git a/src/UserGuide/V1.3.x/User-Manual/User-defined-function_apache.md b/src/UserGuide/V1.3.x/User-Manual/User-defined-function_apache.md index 3b63c81ef..0dbf0cb32 100644 --- a/src/UserGuide/V1.3.x/User-Manual/User-defined-function_apache.md +++ b/src/UserGuide/V1.3.x/User-Manual/User-defined-function_apache.md @@ -1,4 +1,4 @@ -# USER-DEFINED FUNCTION (UDF) +# UDF ## 1. UDF Introduction diff --git a/src/UserGuide/V1.3.x/User-Manual/User-defined-function_timecho.md b/src/UserGuide/V1.3.x/User-Manual/User-defined-function_timecho.md index c8b104e7a..a72e1305f 100644 --- a/src/UserGuide/V1.3.x/User-Manual/User-defined-function_timecho.md +++ b/src/UserGuide/V1.3.x/User-Manual/User-defined-function_timecho.md @@ -1,4 +1,4 @@ -# USER-DEFINED FUNCTION (UDF) +# UDF ## 1. UDF Introduction diff --git a/src/UserGuide/dev-1.3/User-Manual/AINode_apache.md b/src/UserGuide/dev-1.3/AI-capability/AINode_apache.md similarity index 98% rename from src/UserGuide/dev-1.3/User-Manual/AINode_apache.md rename to src/UserGuide/dev-1.3/AI-capability/AINode_apache.md index 3045fa625..95fcccd35 100644 --- a/src/UserGuide/dev-1.3/User-Manual/AINode_apache.md +++ b/src/UserGuide/dev-1.3/AI-capability/AINode_apache.md @@ -19,7 +19,7 @@ --> -# AI Capability（AINode） +# AINode AINode is the third internal node after ConfigNode and DataNode in Apache IoTDB, which extends the capability of machine learning analysis of time series by interacting with DataNode and ConfigNode of IoTDB cluster, supports the introduction of pre-existing machine learning models from the outside to be registered, and uses the registered models in the It supports the process of introducing existing machine learning models from outside for registration, and using the registered models to complete the time series analysis tasks on the specified time series data through simple SQL statements, which integrates the model creation, management and inference in the database engine. At present, we have provided machine learning algorithms or self-developed models for common timing analysis scenarios (e.g. prediction and anomaly detection). @@ -33,7 +33,7 @@ The responsibilities of the three nodes are as follows: - **DataNode**: responsible for receiving and parsing SQL requests from users; responsible for storing time-series data; responsible for preprocessing computation of data. - **AINode**: responsible for model file import creation and model inference. -## Advantageous features +## 1. Advantageous features Compared with building a machine learning service alone, it has the following advantages: @@ -50,7 +50,7 @@ Compared with building a machine learning service alone, it has the following ad -## Basic Concepts +## 2. Basic Concepts - **Model**: a machine learning model that takes time-series data as input and outputs the results or decisions of an analysis task. Model is the basic management unit of AINode, which supports adding (registration), deleting, checking, and using (inference) of models. - **Create**: Load externally designed or trained model files or algorithms into MLNode for unified management and use by IoTDB. @@ -61,16 +61,16 @@ Compared with building a machine learning service alone, it has the following ad :::: -## Installation and Deployment +## 3. Installation and Deployment The deployment of AINode can be found in the document [Deployment Guidelines](../Deployment-and-Maintenance/AINode_Deployment_apache.md#ainode-deployment) . -## Usage Guidelines +## 4. Usage Guidelines AINode provides model creation and deletion process for deep learning models related to timing data. Built-in models do not need to be created and deleted, they can be used directly, and the built-in model instances created after inference is completed will be destroyed automatically. -### Registering Models +### 4.1 Registering Models A trained deep learning model can be registered by specifying the vector dimensions of the model's inputs and outputs, which can be used for model inference. @@ -156,7 +156,7 @@ After the SQL is executed, the registration process will be carried out asynchro Once the model registration is complete, you can call specific functions and perform model inference by using normal queries. -### Viewing Models +### 4.2 Viewing Models Successfully registered models can be queried for model-specific information through the show models command. The SQL definition is as follows: @@ -204,7 +204,7 @@ IoTDB> show models We have registered the corresponding model earlier, you can view the model status through the corresponding designation, active indicates that the model is successfully registered and can be used for inference. -### Delete Model +### 4.3 Delete Model For a successfully registered model, the user can delete it via SQL. In addition to deleting the meta information on the configNode, this operation also deletes all the related model files under the AINode. The SQL is as follows: @@ -214,7 +214,7 @@ drop model You need to specify the model model_name that has been successfully registered to delete the corresponding model. Since model deletion involves the deletion of data on multiple nodes, the operation will not be completed immediately, and the state of the model at this time is DROPPING, and the model in this state cannot be used for model inference. -### Using Built-in Model Reasoning +### 4.4 Using Built-in Model Reasoning The SQL syntax is as follows: @@ -284,7 +284,7 @@ IoTDB> call inference(_Stray, "select s0 from root.eg.airline", k=2) Total line number = 144 ``` -### Reasoning with Deep Learning Models +### 4.5 Reasoning with Deep Learning Models The SQL syntax is as follows: @@ -444,7 +444,7 @@ Total line number = 4 In the result set, each row's label corresponds to the output of the anomaly detection model after inputting each group of 24 rows of data. -## Privilege Management +## 5. Privilege Management When using AINode related functions, the authentication of IoTDB itself can be used to do a permission management, users can only use the model management related functions when they have the USE_MODEL permission. When using the inference function, the user needs to have the permission to access the source sequence corresponding to the SQL of the input model. @@ -453,9 +453,9 @@ When using AINode related functions, the authentication of IoTDB itself can be u | USE_MODEL | create model/show models/drop model | √ | √ | x | | READ_DATA| call inference | √ | √|√ | -## Practical Examples +## 6. Practical Examples -### Power Load Prediction +### 6.1 Power Load Prediction In some industrial scenarios, there is a need to predict power loads, which can be used to optimise power supply, conserve energy and resources, support planning and expansion, and enhance power system reliability. @@ -526,7 +526,7 @@ The data before 10/24 00:00 represents the past data input to the model, the blu As can be seen, we have used the relationship between the six load information and the corresponding time oil temperatures for the past 96 hours (4 days) to model the possible changes in this data for the oil temperature for the next 48 hours (2 days) based on the inter-relationships between the sequences learned previously, and it can be seen that the predicted curves maintain a high degree of consistency in trend with the actual results after visualisation. -### Power Prediction +### 6.2 Power Prediction Power monitoring of current, voltage and power data is required in substations for detecting potential grid problems, identifying faults in the power system, effectively managing grid loads and analysing power system performance and trends. @@ -592,7 +592,7 @@ The data before 02/14 20:48 represents the past data input to the model, the blu It can be seen that we used the voltage data from the past 10 minutes and, based on the previously learned inter-sequence relationships, modeled the possible changes in the phase C voltage data for the next 5 minutes. The visualized forecast curve shows a certain degree of synchronicity with the actual results in terms of trend. -### Anomaly Detection +### 6.3 Anomaly Detection In the civil aviation and transport industry, there exists a need for anomaly detection of the number of passengers travelling on an aircraft. The results of anomaly detection can be used to guide the adjustment of flight scheduling to make the organisation more efficient. diff --git a/src/UserGuide/dev-1.3/User-Manual/AINode_timecho.md b/src/UserGuide/dev-1.3/AI-capability/AINode_timecho.md similarity index 98% rename from src/UserGuide/dev-1.3/User-Manual/AINode_timecho.md rename to src/UserGuide/dev-1.3/AI-capability/AINode_timecho.md index b797c8dfb..204f91cf9 100644 --- a/src/UserGuide/dev-1.3/User-Manual/AINode_timecho.md +++ b/src/UserGuide/dev-1.3/AI-capability/AINode_timecho.md @@ -19,7 +19,7 @@ --> -# AI Capability（AINode） +# AINode AINode is the third internal node after ConfigNode and DataNode in Apache IoTDB, which extends the capability of machine learning analysis of time series by interacting with DataNode and ConfigNode of IoTDB cluster, supports the introduction of pre-existing machine learning models from the outside to be registered, and uses the registered models in the It supports the process of introducing existing machine learning models from outside for registration, and using the registered models to complete the time series analysis tasks on the specified time series data through simple SQL statements, which integrates the model creation, management and inference in the database engine. At present, we have provided machine learning algorithms or self-developed models for common timing analysis scenarios (e.g. prediction and anomaly detection). @@ -33,7 +33,7 @@ The responsibilities of the three nodes are as follows: - **DataNode**: responsible for receiving and parsing SQL requests from users; responsible for storing time-series data; responsible for preprocessing computation of data. - **AINode**: responsible for model file import creation and model inference. -## Advantageous features +## 1. Advantageous features Compared with building a machine learning service alone, it has the following advantages: @@ -50,7 +50,7 @@ Compared with building a machine learning service alone, it has the following ad -## Basic Concepts +## 2. Basic Concepts - **Model**: a machine learning model that takes time-series data as input and outputs the results or decisions of an analysis task. Model is the basic management unit of AINode, which supports adding (registration), deleting, checking, and using (inference) of models. - **Create**: Load externally designed or trained model files or algorithms into MLNode for unified management and use by IoTDB. @@ -61,16 +61,16 @@ Compared with building a machine learning service alone, it has the following ad :::: -## Installation and Deployment +## 3. Installation and Deployment The deployment of AINode can be found in the document [Deployment Guidelines](../Deployment-and-Maintenance/AINode_Deployment_timecho.md#AINode-部署) . -## Usage Guidelines +## 4. Usage Guidelines AINode provides model creation and deletion process for deep learning models related to timing data. Built-in models do not need to be created and deleted, they can be used directly, and the built-in model instances created after inference is completed will be destroyed automatically. -### Registering Models +### 4.1 Registering Models A trained deep learning model can be registered by specifying the vector dimensions of the model's inputs and outputs, which can be used for model inference. @@ -156,7 +156,7 @@ After the SQL is executed, the registration process will be carried out asynchro Once the model registration is complete, you can call specific functions and perform model inference by using normal queries. -### Viewing Models +### 4.2 Viewing Models Successfully registered models can be queried for model-specific information through the show models command. The SQL definition is as follows: @@ -204,7 +204,7 @@ IoTDB> show models We have registered the corresponding model earlier, you can view the model status through the corresponding designation, active indicates that the model is successfully registered and can be used for inference. -### Delete Model +### 4.3 Delete Model For a successfully registered model, the user can delete it via SQL. In addition to deleting the meta information on the configNode, this operation also deletes all the related model files under the AINode. The SQL is as follows: @@ -214,7 +214,7 @@ drop model You need to specify the model model_name that has been successfully registered to delete the corresponding model. Since model deletion involves the deletion of data on multiple nodes, the operation will not be completed immediately, and the state of the model at this time is DROPPING, and the model in this state cannot be used for model inference. -### Using Built-in Model Reasoning +### 4.4 Using Built-in Model Reasoning The SQL syntax is as follows: @@ -284,7 +284,7 @@ IoTDB> call inference(_Stray, "select s0 from root.eg.airline", k=2) Total line number = 144 ``` -### Reasoning with Deep Learning Models +### 4.5 Reasoning with Deep Learning Models The SQL syntax is as follows: @@ -444,7 +444,13 @@ Total line number = 4 In the result set, each row's label corresponds to the output of the anomaly detection model after inputting each group of 24 rows of data. -## Privilege Management + +### 4.6 TimeSeries Large Models Import Steps + +AINode currently supports a variety of time series large models. For deployment and usage, please refer to [TimeSeries Large Models](../AI-capability/TimeSeries-Large-Model) + + +## 5. Privilege Management When using AINode related functions, the authentication of IoTDB itself can be used to do a permission management, users can only use the model management related functions when they have the USE_MODEL permission. When using the inference function, the user needs to have the permission to access the source sequence corresponding to the SQL of the input model. @@ -453,9 +459,9 @@ When using AINode related functions, the authentication of IoTDB itself can be u | USE_MODEL | create model/show models/drop model | √ | √ | x | | READ_DATA| call inference | √ | √|√ | -## Practical Examples +## 6. Practical Examples -### Power Load Prediction +### 6.1 Power Load Prediction In some industrial scenarios, there is a need to predict power loads, which can be used to optimise power supply, conserve energy and resources, support planning and expansion, and enhance power system reliability. @@ -526,7 +532,7 @@ The data before 10/24 00:00 represents the past data input to the model, the blu As can be seen, we have used the relationship between the six load information and the corresponding time oil temperatures for the past 96 hours (4 days) to model the possible changes in this data for the oil temperature for the next 48 hours (2 days) based on the inter-relationships between the sequences learned previously, and it can be seen that the predicted curves maintain a high degree of consistency in trend with the actual results after visualisation. -### Power Prediction +### 6.2 Power Prediction Power monitoring of current, voltage and power data is required in substations for detecting potential grid problems, identifying faults in the power system, effectively managing grid loads and analysing power system performance and trends. @@ -592,7 +598,7 @@ The data before 02/14 20:48 represents the past data input to the model, the blu It can be seen that we used the voltage data from the past 10 minutes and, based on the previously learned inter-sequence relationships, modeled the possible changes in the phase C voltage data for the next 5 minutes. The visualized forecast curve shows a certain degree of synchronicity with the actual results in terms of trend. -### Anomaly Detection +### 6.3 Anomaly Detection In the civil aviation and transport industry, there exists a need for anomaly detection of the number of passengers travelling on an aircraft. The results of anomaly detection can be used to guide the adjustment of flight scheduling to make the organisation more efficient. diff --git a/src/UserGuide/dev-1.3/AI-capability/TimeSeries-Large-Model.md b/src/UserGuide/dev-1.3/AI-capability/TimeSeries-Large-Model.md new file mode 100644 index 000000000..4ce2801d0 --- /dev/null +++ b/src/UserGuide/dev-1.3/AI-capability/TimeSeries-Large-Model.md @@ -0,0 +1,117 @@ + + +# TimeSeries Large Model + +## Introduction + +A time series large model is a foundational model specifically designed for time series analysis. The IoTDB team has independently developed time series large models, which are pre-trained on massive time series data using technologies such as transformer structures. These models can understand and generate time series data across various domains and are applicable to applications like time series forecasting, anomaly detection, and time series imputation. Unlike traditional time series analysis techniques, time series large models possess the capability to extract universal features and provide technical services based on zero-shot analysis and fine-tuning for a wide range of analytical tasks. + +The team's related technologies of time series large models have been published in top international machine learning conferences. + +## Application Scenarios + +- **Time Series Forecasting**: Provides forecasting services for time series data in industrial production, natural environments, and other areas, helping users to understand future trends in advance. +- **Data Imputation**: For missing segments in time series, perform context imputation to enhance the continuity and completeness of the dataset. +- **Anomaly Detection**: Utilizing regression analysis technology, monitor time series data in real-time and provide timely warnings for potential anomalies. + +![](/img/LargeModel10.png) + +## Timer Model + +The Timer model not only demonstrates excellent few-shot generalization and multi-task adaptation capabilities but also gains a rich knowledge base through pre-training, endowing it with the universal capability to handle a variety of downstream tasks, featuring the following: + +- **Generalization**: The model can be fine-tuned using a small number of samples to achieve leading predictive performance in the industry. +- **Versatility**: The model is designed flexibly to adapt to various task requirements and supports variable input and output lengths, enabling it to play a role in various application scenarios. +- **Scalability**: As the number of model parameters increases or the scale of pre-training data expands, the model's performance continues to improve, ensuring the model can optimize its predictive effects with the growth of time and data volume. + +![](/img/LargeModel02.png) + +## Timer-XL Model + +Timer-XL is an upgraded version of Timer that further extends the network structure and achieves comprehensive breakthroughs in multiple dimensions: + +- **Long Context Support**: This model breaks through the limitations of traditional time series forecasting models, supporting the processing of thousands of tokens (equivalent to tens of thousands of time points) of input, effectively addressing the bottleneck of context length. +- **Multi-variable Forecasting Scenario Coverage**: Supports a variety of forecasting scenarios, including non-stationary time series forecasting, multi-variable prediction tasks, and predictions involving covariates, meeting diverse business needs. +- **Large-scale Industrial Time Series Dataset**: Pre-trained using a massive industrial IoT time series dataset that has a large volume, excellent quality, and rich domain characteristics, covering energy, aerospace, steel, transportation, and more. + + +## Effect Demonstration + +Time series large models can adapt to real time series data from various fields and scenarios, showing excellent processing effects in various tasks. Here are the real performances on different data: + +**Time Series Forecasting:** + +Utilizing the predictive capabilities of the time series large model, it can accurately predict the future trend of time series. As shown in the figure, the blue curve represents the predicted trend, and the red curve represents the actual trend, with the two curves highly matching. + +![](/img/LargeModel03.png) + +**Data Imputation:**： + +Using the time series large model to perform predictive imputation for missing data segments. + +![](/img/LargeModel04.png) + + +**Anomaly Detection:**： + +Utilizing the time series large model to accurately identify anomalies that deviate significantly from the normal trend. + +![](/img/LargeModel05.png) + +## Deployment Usage + +1. Open the IoTDB CLI console and verify that the ConfigNode, DataNode, and AINode statuses are all ​Running. + +Check command: + +```sql +show cluster +``` + +![](/img/ainode-timer-1.png) + +2. Model file storage path: It is recommended to place the model files in the same directory as the AINode installation package. + You may create a new folder to store model files. + +3. Register the model + +Use the following SQL statement: + +```sql +create model using uri +``` + +Example (for the Timer model): + +```sql +create model Timer using uri +``` + +4. Verify model registration success + +Check command: + +```sql +show models +``` + +![](/img/LargeModel06.png) diff --git a/src/UserGuide/dev-1.3/User-Manual/Audit-Log_timecho.md b/src/UserGuide/dev-1.3/User-Manual/Audit-Log_timecho.md index d3ff3aae6..976b7da17 100644 --- a/src/UserGuide/dev-1.3/User-Manual/Audit-Log_timecho.md +++ b/src/UserGuide/dev-1.3/User-Manual/Audit-Log_timecho.md @@ -19,7 +19,7 @@ --> -# Audit log +# Security Audit ## Background of the function diff --git a/src/UserGuide/dev-1.3/User-Manual/Streaming_apache.md b/src/UserGuide/dev-1.3/User-Manual/Streaming_apache.md index cfb6a3433..7f9b9716a 100644 --- a/src/UserGuide/dev-1.3/User-Manual/Streaming_apache.md +++ b/src/UserGuide/dev-1.3/User-Manual/Streaming_apache.md @@ -19,7 +19,7 @@ --> -# Stream Processing +# Stream Computing Framework The IoTDB stream processing framework allows users to implement customized stream processing logic, which can monitor and capture storage engine changes, transform changed data, and push transformed data outward. diff --git a/src/UserGuide/dev-1.3/User-Manual/Streaming_timecho.md b/src/UserGuide/dev-1.3/User-Manual/Streaming_timecho.md index 8d6f50e44..80edebe9c 100644 --- a/src/UserGuide/dev-1.3/User-Manual/Streaming_timecho.md +++ b/src/UserGuide/dev-1.3/User-Manual/Streaming_timecho.md @@ -19,7 +19,7 @@ --> -# Stream Processing +# Stream Computing Framework The IoTDB stream processing framework allows users to implement customized stream processing logic, which can monitor and capture storage engine changes, transform changed data, and push transformed data outward. diff --git a/src/UserGuide/dev-1.3/User-Manual/User-defined-function_apache.md b/src/UserGuide/dev-1.3/User-Manual/User-defined-function_apache.md index 27a9b2eec..172734b8a 100644 --- a/src/UserGuide/dev-1.3/User-Manual/User-defined-function_apache.md +++ b/src/UserGuide/dev-1.3/User-Manual/User-defined-function_apache.md @@ -1,4 +1,4 @@ -# USER-DEFINED FUNCTION (UDF) +# UDF ## 1. UDF Introduction diff --git a/src/UserGuide/dev-1.3/User-Manual/User-defined-function_timecho.md b/src/UserGuide/dev-1.3/User-Manual/User-defined-function_timecho.md index ed20643e6..32dc34b3f 100644 --- a/src/UserGuide/dev-1.3/User-Manual/User-defined-function_timecho.md +++ b/src/UserGuide/dev-1.3/User-Manual/User-defined-function_timecho.md @@ -1,4 +1,4 @@ -# USER-DEFINED FUNCTION (UDF) +# UDF ## 1. UDF Introduction diff --git a/src/UserGuide/latest/User-Manual/Audit-Log_timecho.md b/src/UserGuide/latest/User-Manual/Audit-Log_timecho.md index 77faf0a7a..61ba509c5 100644 --- a/src/UserGuide/latest/User-Manual/Audit-Log_timecho.md +++ b/src/UserGuide/latest/User-Manual/Audit-Log_timecho.md @@ -19,7 +19,7 @@ --> -# Audit log +# Security Audit ## 1. Background of the function diff --git a/src/UserGuide/latest/User-Manual/Streaming_apache.md b/src/UserGuide/latest/User-Manual/Streaming_apache.md index da81e199f..b5a8e2419 100644 --- a/src/UserGuide/latest/User-Manual/Streaming_apache.md +++ b/src/UserGuide/latest/User-Manual/Streaming_apache.md @@ -19,7 +19,7 @@ --> -# Stream Processing +# Stream Computing Framework The IoTDB stream processing framework allows users to implement customized stream processing logic, which can monitor and capture storage engine changes, transform changed data, and push transformed data outward. diff --git a/src/UserGuide/latest/User-Manual/Streaming_timecho.md b/src/UserGuide/latest/User-Manual/Streaming_timecho.md index 3f72c1f8f..07a3e3017 100644 --- a/src/UserGuide/latest/User-Manual/Streaming_timecho.md +++ b/src/UserGuide/latest/User-Manual/Streaming_timecho.md @@ -19,7 +19,7 @@ --> -# Stream Processing +# Stream Computing Framework The IoTDB stream processing framework allows users to implement customized stream processing logic, which can monitor and capture storage engine changes, transform changed data, and push transformed data outward. diff --git a/src/zh/UserGuide/Master/Tree/User-Manual/Audit-Log_timecho.md b/src/zh/UserGuide/Master/Tree/User-Manual/Audit-Log_timecho.md index 4f88f5732..3fd2eac0d 100644 --- a/src/zh/UserGuide/Master/Tree/User-Manual/Audit-Log_timecho.md +++ b/src/zh/UserGuide/Master/Tree/User-Manual/Audit-Log_timecho.md @@ -20,7 +20,7 @@ --> -# 审计日志 +# 安全审计 ## 1. 功能背景 diff --git a/src/zh/UserGuide/Master/Tree/User-Manual/Streaming_apache.md b/src/zh/UserGuide/Master/Tree/User-Manual/Streaming_apache.md index 8e7fd3e58..fa974851d 100644 --- a/src/zh/UserGuide/Master/Tree/User-Manual/Streaming_apache.md +++ b/src/zh/UserGuide/Master/Tree/User-Manual/Streaming_apache.md @@ -19,7 +19,7 @@ --> -# 流处理框架 +# 流计算框架 IoTDB 流处理框架允许用户实现自定义的流处理逻辑，可以实现对存储引擎变更的监听和捕获、实现对变更数据的变形、实现对变形后数据的向外推送等逻辑。 diff --git a/src/zh/UserGuide/Master/Tree/User-Manual/Streaming_timecho.md b/src/zh/UserGuide/Master/Tree/User-Manual/Streaming_timecho.md index b8ecd936e..a71bc7361 100644 --- a/src/zh/UserGuide/Master/Tree/User-Manual/Streaming_timecho.md +++ b/src/zh/UserGuide/Master/Tree/User-Manual/Streaming_timecho.md @@ -19,7 +19,7 @@ --> -# IoTDB 流处理框架 +# 流计算框架 IoTDB 流处理框架允许用户实现自定义的流处理逻辑，可以实现对存储引擎变更的监听和捕获、实现对变更数据的变形、实现对变形后数据的向外推送等逻辑。 diff --git a/src/zh/UserGuide/Master/Tree/User-Manual/User-defined-function_apache.md b/src/zh/UserGuide/Master/Tree/User-Manual/User-defined-function_apache.md index fb4489c66..25d78f7aa 100644 --- a/src/zh/UserGuide/Master/Tree/User-Manual/User-defined-function_apache.md +++ b/src/zh/UserGuide/Master/Tree/User-Manual/User-defined-function_apache.md @@ -1,4 +1,4 @@ -# 用户自定义函数 +# UDF ## 1. UDF 介绍 diff --git a/src/zh/UserGuide/Master/Tree/User-Manual/User-defined-function_timecho.md b/src/zh/UserGuide/Master/Tree/User-Manual/User-defined-function_timecho.md index fa62995b4..38da4c2e5 100644 --- a/src/zh/UserGuide/Master/Tree/User-Manual/User-defined-function_timecho.md +++ b/src/zh/UserGuide/Master/Tree/User-Manual/User-defined-function_timecho.md @@ -1,4 +1,4 @@ -# 用户自定义函数 +# UDF ## 1. UDF 介绍 diff --git a/src/zh/UserGuide/V1.2.x/User-Manual/Security-Management_timecho.md b/src/zh/UserGuide/V1.2.x/User-Manual/Security-Management_timecho.md index c7a02b3d3..48b84b8ad 100644 --- a/src/zh/UserGuide/V1.2.x/User-Manual/Security-Management_timecho.md +++ b/src/zh/UserGuide/V1.2.x/User-Manual/Security-Management_timecho.md @@ -69,7 +69,7 @@ white.list: ![白名单](/img/%E7%99%BD%E5%90%8D%E5%8D%95.png) -## 审计日志 +## 安全审计 ### 功能背景 diff --git a/src/zh/UserGuide/V1.2.x/User-Manual/Streaming.md b/src/zh/UserGuide/V1.2.x/User-Manual/Streaming.md index db8960d7e..7d8c36b98 100644 --- a/src/zh/UserGuide/V1.2.x/User-Manual/Streaming.md +++ b/src/zh/UserGuide/V1.2.x/User-Manual/Streaming.md @@ -19,7 +19,7 @@ --> -# 流处理框架 +# 流计算框架 IoTDB 流处理框架允许用户实现自定义的流处理逻辑，可以实现对存储引擎变更的监听和捕获、实现对变更数据的变形、实现对变形后数据的向外推送等逻辑。 diff --git a/src/zh/UserGuide/V1.2.x/User-Manual/Streaming_timecho.md b/src/zh/UserGuide/V1.2.x/User-Manual/Streaming_timecho.md index a613957da..1d4e8e6aa 100644 --- a/src/zh/UserGuide/V1.2.x/User-Manual/Streaming_timecho.md +++ b/src/zh/UserGuide/V1.2.x/User-Manual/Streaming_timecho.md @@ -19,7 +19,7 @@ --> -# 流处理框架 +# 流计算框架 IoTDB 流处理框架允许用户实现自定义的流处理逻辑，可以实现对存储引擎变更的监听和捕获、实现对变更数据的变形、实现对变形后数据的向外推送等逻辑。 diff --git a/src/zh/UserGuide/V1.3.x/User-Manual/AINode_apache.md b/src/zh/UserGuide/V1.3.x/AI-capability/AINode_apache.md similarity index 99% rename from src/zh/UserGuide/V1.3.x/User-Manual/AINode_apache.md rename to src/zh/UserGuide/V1.3.x/AI-capability/AINode_apache.md index 6ff0a959f..b6620803f 100644 --- a/src/zh/UserGuide/V1.3.x/User-Manual/AINode_apache.md +++ b/src/zh/UserGuide/V1.3.x/AI-capability/AINode_apache.md @@ -19,7 +19,7 @@ --> -# AI能力（AINode） +# AINode AINode 是 IoTDB 在ConfigNode、DataNode后提供的第三种内生节点，该节点通过与 IoTDB 集群的 DataNode、ConfigNode 的交互，扩展了对时间序列进行机器学习分析的能力，支持从外部引入已有机器学习模型进行注册，并使用注册的模型在指定时序数据上通过简单 SQL 语句完成时序分析任务的过程，将模型的创建、管理及推理融合在数据库引擎中。目前已提供常见时序分析场景（例如预测与异常检测）的机器学习算法或自研模型。 @@ -33,7 +33,7 @@ AINode 是 IoTDB 在ConfigNode、DataNode后提供的第三种内生节点，该 - **DataNode**：负责接收并解析用户的 SQL请求；负责存储时间序列数据；负责数据的预处理计算。 - **AINode**：负责模型文件的导入创建以及模型推理。 -## 优势特点 +## 1. 优势特点 与单独构建机器学习服务相比，具有以下优势： @@ -49,7 +49,7 @@ AINode 是 IoTDB 在ConfigNode、DataNode后提供的第三种内生节点，该 - **时间序列标注（Time Series Annotation）**：为每个数据点或特定时间段添加额外的信息或标记，例如事件发生、异常点、趋势变化等，以便更好地理解和分析数据。 -## 基本概念 +## 2. 基本概念 - **模型（Model）**：机器学习模型，以时序数据作为输入，输出分析任务的结果或决策。模型是AINode 的基本管理单元，支持模型的增（注册）、删、查、用（推理）。 - **创建（Create）**: 将外部设计或训练好的模型文件或算法加载到MLNode中，由IoTDB统一管理与使用。 @@ -60,15 +60,15 @@ AINode 是 IoTDB 在ConfigNode、DataNode后提供的第三种内生节点，该 ::: -## 安装部署 +## 3. 安装部署 AINode 的部署可参考文档 [部署指导](../Deployment-and-Maintenance/AINode_Deployment_apache.md#ainode-部署) 章节。 -## 使用指导 +## 4. 使用指导 AINode 对时序数据相关的深度学习模型提供了模型创建及删除的流程，内置模型无需创建及删除，可直接使用，并且在完成推理后创建的内置模型实例将自动销毁。 -### 注册模型 +### 4.1 注册模型 通过指定模型输入输出的向量维度，可以注册训练好的深度学习模型，从而用于模型推理。 @@ -154,7 +154,7 @@ SQL执行后会异步进行注册的流程，可以通过模型展示查看模 模型注册完成后，就可以通过使用正常查询的方式调用具体函数，进行模型推理。 -### 查看模型 +### 4.2 查看模型 注册成功的模型可以通过show models指令查询模型的具体信息。其SQL定义如下： @@ -202,7 +202,7 @@ IoTDB> show models 我们前面已经注册了对应的模型，可以通过对应的指定查看模型状态，active表明模型注册成功，可用于推理。 -### 删除模型 +### 4.3 删除模型 对于注册成功的模型，用户可以通过SQL进行删除。该操作除了删除configNode上的元信息外，还会删除所有AINode下的相关模型文件。其SQL如下： @@ -212,7 +212,7 @@ drop model 需要指定已经成功注册的模型model_name来删除对应的模型。由于模型删除涉及多个节点上的数据删除，操作不会立即完成，此时模型的状态为DROPPING，该状态的模型不能用于模型推理。 -### 使用内置模型推理 +### 4.4 使用内置模型推理 SQL语法如下： @@ -281,7 +281,7 @@ IoTDB> call inference(_Stray, "select s0 from root.eg.airline", k=2) Total line number = 144 ``` -### 使用深度学习模型推理 +### 4.5 使用深度学习模型推理 SQL语法如下： @@ -441,7 +441,7 @@ Total line number = 4 其中结果集中每行的标签对应每24行数据为一组，输入该异常检测模型后的输出。 -## 权限管理 +## 5. 权限管理 使用AINode相关的功能时，可以使用IoTDB本身的鉴权去做一个权限管理，用户只有在具备 USE_MODEL 权限时，才可以使用模型管理的相关功能。当使用推理功能时，用户需要有访问输入模型的SQL对应的源序列的权限。 @@ -450,9 +450,9 @@ Total line number = 4 | USE_MODEL | create model / show models / drop model | √ | √ | x | | READ_DATA | call inference | √ | √ | √ | -## 实际案例 +## 6. 实际案例 -### 电力负载预测 +### 6.1 电力负载预测 在部分工业场景下，会存在预测电力负载的需求，预测结果可用于优化电力供应、节约能源和资源、支持规划和扩展以及增强电力系统的可靠性。 @@ -523,7 +523,7 @@ Total line number = 48 可以看到，我们使用了过去96个小时（4天）的六个负载信息和对应时间油温的关系，基于之前学习到的序列间相互关系对未来48个小时（2天）的油温这一数据的可能变化进行了建模，可以看到可视化后预测曲线与实际结果在趋势上保持了较高程度的一致性。 -### 功率预测 +### 6.2 功率预测 变电站需要对电流、电压、功率等数据进行电力监控，用于检测潜在的电网问题、识别电力系统中的故障、有效管理电网负载以及分析电力系统的性能和趋势等。 @@ -588,7 +588,7 @@ Total line number = 48 可以看到，我们使用了过去10分钟的电压的数据，基于之前学习到的序列间相互关系对未来5分钟的C相电压这一数据的可能变化进行了建模，可以看到可视化后预测曲线与实际结果在趋势上保持了一定的同步性。 -### 异常检测 +### 6.3 异常检测 在民航交通运输业，存在着对乘机旅客数量进行异常检测的需求。异常检测的结果可用于指导调整航班的调度，以使得企业获得更大效益。 diff --git a/src/zh/UserGuide/V1.3.x/User-Manual/AINode_timecho.md b/src/zh/UserGuide/V1.3.x/AI-capability/AINode_timecho.md similarity index 98% rename from src/zh/UserGuide/V1.3.x/User-Manual/AINode_timecho.md rename to src/zh/UserGuide/V1.3.x/AI-capability/AINode_timecho.md index 594236294..c67c41de5 100644 --- a/src/zh/UserGuide/V1.3.x/User-Manual/AINode_timecho.md +++ b/src/zh/UserGuide/V1.3.x/AI-capability/AINode_timecho.md @@ -19,7 +19,7 @@ --> -# AI能力（AINode） +# AINode AINode 是 IoTDB 在ConfigNode、DataNode后提供的第三种内生节点，该节点通过与 IoTDB 集群的 DataNode、ConfigNode 的交互，扩展了对时间序列进行机器学习分析的能力，支持从外部引入已有机器学习模型进行注册，并使用注册的模型在指定时序数据上通过简单 SQL 语句完成时序分析任务的过程，将模型的创建、管理及推理融合在数据库引擎中。目前已提供常见时序分析场景（例如预测与异常检测）的机器学习算法或自研模型。 @@ -33,7 +33,7 @@ AINode 是 IoTDB 在ConfigNode、DataNode后提供的第三种内生节点，该 - **DataNode**：负责接收并解析用户的 SQL请求；负责存储时间序列数据；负责数据的预处理计算。 - **AINode**：负责模型文件的导入创建以及模型推理。 -## 优势特点 +## 1. 优势特点 与单独构建机器学习服务相比，具有以下优势： @@ -49,7 +49,7 @@ AINode 是 IoTDB 在ConfigNode、DataNode后提供的第三种内生节点，该 - **时间序列标注（Time Series Annotation）**：为每个数据点或特定时间段添加额外的信息或标记，例如事件发生、异常点、趋势变化等，以便更好地理解和分析数据。 -## 基本概念 +## 2. 基本概念 - **模型（Model）**：机器学习模型，以时序数据作为输入，输出分析任务的结果或决策。模型是AINode 的基本管理单元，支持模型的增（注册）、删、查、用（推理）。 - **创建（Create）**: 将外部设计或训练好的模型文件或算法加载到MLNode中，由IoTDB统一管理与使用。 @@ -60,15 +60,15 @@ AINode 是 IoTDB 在ConfigNode、DataNode后提供的第三种内生节点，该 ::: -## 安装部署 +## 3. 安装部署 AINode 的部署可参考文档 [部署指导](../Deployment-and-Maintenance/AINode_Deployment_timecho.md#AINode-部署) 章节。 -## 使用指导 +## 4. 使用指导 AINode 对时序数据相关的深度学习模型提供了模型创建及删除的流程，内置模型无需创建及删除，可直接使用，并且在完成推理后创建的内置模型实例将自动销毁。 -### 注册模型 +### 4.1 注册模型 通过指定模型输入输出的向量维度，可以注册训练好的深度学习模型，从而用于模型推理。 @@ -154,7 +154,7 @@ SQL执行后会异步进行注册的流程，可以通过模型展示查看模 模型注册完成后，就可以通过使用正常查询的方式调用具体函数，进行模型推理。 -### 查看模型 +### 4.2 查看模型 注册成功的模型可以通过show models指令查询模型的具体信息。其SQL定义如下： @@ -202,7 +202,7 @@ IoTDB> show models 我们前面已经注册了对应的模型，可以通过对应的指定查看模型状态，active表明模型注册成功，可用于推理。 -### 删除模型 +### 4.3 删除模型 对于注册成功的模型，用户可以通过SQL进行删除。该操作除了删除configNode上的元信息外，还会删除所有AINode下的相关模型文件。其SQL如下： @@ -212,7 +212,7 @@ drop model 需要指定已经成功注册的模型model_name来删除对应的模型。由于模型删除涉及多个节点上的数据删除，操作不会立即完成，此时模型的状态为DROPPING，该状态的模型不能用于模型推理。 -### 使用内置模型推理 +### 4.4 使用内置模型推理 SQL语法如下： @@ -281,7 +281,7 @@ IoTDB> call inference(_Stray, "select s0 from root.eg.airline", k=2) Total line number = 144 ``` -### 使用深度学习模型推理 +### 4.5 使用深度学习模型推理 SQL语法如下： @@ -441,7 +441,12 @@ Total line number = 4 其中结果集中每行的标签对应每24行数据为一组，输入该异常检测模型后的输出。 -## 权限管理 + +### 4.6 时序大模型导入步骤 + +AINode 目前支持多种时序大模型，部署使用请参考[时序大模型](../AI-capability/TimeSeries-Large-Model) + +## 5. 权限管理 使用AINode相关的功能时，可以使用IoTDB本身的鉴权去做一个权限管理，用户只有在具备 USE_MODEL 权限时，才可以使用模型管理的相关功能。当使用推理功能时，用户需要有访问输入模型的SQL对应的源序列的权限。 @@ -450,9 +455,9 @@ Total line number = 4 | USE_MODEL | create model / show models / drop model | √ | √ | x | | READ_DATA | call inference | √ | √ | √ | -## 实际案例 +## 6. 实际案例 -### 电力负载预测 +### 6.1 电力负载预测 在部分工业场景下，会存在预测电力负载的需求，预测结果可用于优化电力供应、节约能源和资源、支持规划和扩展以及增强电力系统的可靠性。 @@ -523,7 +528,7 @@ Total line number = 48 可以看到，我们使用了过去96个小时（4天）的六个负载信息和对应时间油温的关系，基于之前学习到的序列间相互关系对未来48个小时（2天）的油温这一数据的可能变化进行了建模，可以看到可视化后预测曲线与实际结果在趋势上保持了较高程度的一致性。 -### 功率预测 +### 6.2 功率预测 变电站需要对电流、电压、功率等数据进行电力监控，用于检测潜在的电网问题、识别电力系统中的故障、有效管理电网负载以及分析电力系统的性能和趋势等。 @@ -588,7 +593,7 @@ Total line number = 48 可以看到，我们使用了过去10分钟的电压的数据，基于之前学习到的序列间相互关系对未来5分钟的C相电压这一数据的可能变化进行了建模，可以看到可视化后预测曲线与实际结果在趋势上保持了一定的同步性。 -### 异常检测 +### 6.3 异常检测 在民航交通运输业，存在着对乘机旅客数量进行异常检测的需求。异常检测的结果可用于指导调整航班的调度，以使得企业获得更大效益。 diff --git a/src/zh/UserGuide/V1.3.x/AI-capability/TimeSeries-Large-Model.md b/src/zh/UserGuide/V1.3.x/AI-capability/TimeSeries-Large-Model.md new file mode 100644 index 000000000..cf7804394 --- /dev/null +++ b/src/zh/UserGuide/V1.3.x/AI-capability/TimeSeries-Large-Model.md @@ -0,0 +1,111 @@ + + +# 时序大模型 + +## 简介 + +时序大模型是一种专为时序数据分析设计的基础模型。IoTDB 团队长期自研时序大模型，基于变换器（Transformer）结构等技术在海量时序数据上预训练，能够理解并生成多种领域的时序数据，可被应用于时序预测、异常检测、时序填补等应用场景。不同于传统时序分析技术，时序大模型具备通用特征提取能力，基于零样本分析、微调等技术服务广泛的分析任务。 + +团队所研时序大模型相关技术均发表在国际机器学习顶级会议。 + +## 应用场景 + +- **时序预测**：为工业生产、自然环境等领域提供时间序列数据的预测服务，帮助用户提前了解未来趋势。 +- **数据填补**：针对时间序列中的缺失序列段，进行上下文填补，以增强数据集的连续性和完整性。 +- **异常检测**：利用自回归分析技术，对时间序列数据进行实时监测，及时预警潜在的异常情况。 + +![](/img/LargeModel09.png) + +## Timer 模型 + +Timer模型不仅展现了出色的少样本泛化和多任务适配能力，还通过预训练获得了丰富的知识库，赋予了它处理多样化下游任务的通用能力，拥有以下特点： + +- **泛化性**：模型能够通过使用少量样本进行微调，达到行业内领先的深度模型预测效果。 +- **通用性**：模型设计灵活，能够适配多种不同的任务需求，并且支持变化的输入和输出长度，使其在各种应用场景中都能发挥作用。 +- **可扩展性**：随着模型参数数量的增加或预训练数据规模的扩大，模型的性能会持续提升，确保模型能够随着时间和数据量的增长而不断优化其预测效果。 + +![](/img/LargeModel02.png) + +## Timer-XL 模型 + +Timer-XL 基于 Timer 进一步扩展升级了网络结构，在多个维度上进行全面突破： + +- **超长上下文支持**：该模型突破了传统时序预测模型的限制，支持处理数千个Token（相当于数万个时间点）的输入，有效解决了上下文长度的瓶颈问题。 +- **多变量预测场景覆盖**：支持多种预测场景，包括非平稳时间序列的预测、涉及多个变量的预测任务以及包含协变量的预测，满足多样化的业务需求。 +- **大规模工业时序数据集：**采用万亿大规模工业物联网领域的时序数据集进行预训练，数据集兼有庞大的体量、卓越的质量和丰富的领域等重要特质，覆盖能源、航空航天、钢铁、交通等多领域。 + +## 效果展示 + +时序大模型能够适应多种不同领域和场景的真实时序数据，在各种任务上拥有优异的处理效果，以下是在不同数据上的真实表现： + +**时序预测：** + +利用时序大模型的预测能力，能够准确预测时间序列的未来变化趋势，如下图蓝色曲线代表预测趋势，红色曲线为实际趋势，两曲线高度吻合。 + +![](/img/LargeModel03.png) + +**数据填补：**： + +利用时序大模型对缺失数据段进行预测式填补。 + +![](/img/LargeModel04.png) + + +**异常检测：**： + +利用时序大模型精准识别与正常趋势偏离过大的异常值。 + +![](/img/LargeModel05.png) + +## 部署使用 + +1. 打开 IoTDB cli 控制台，检查 ConfigNode、DataNode、AINode 节点确保均为 Running。 + +检查命令： +```sql +show cluster +``` + +![](/img/ainode-timer-1.png) + +2. 模型文件存放路径：推荐放在 AINode 安装包相同文件夹下，可新建模型文件夹存放模型文件 +3. 注册模型语句 + +```sql +create model using uri +``` + +示例： + +```sql +create model Timer-xl using uri +``` + +4. 检查模型是否注册成功 + +检查命令： + +```sql +show models +``` + +![](/img/LargeModel06.png) diff --git a/src/zh/UserGuide/V1.3.x/User-Manual/Audit-Log_timecho.md b/src/zh/UserGuide/V1.3.x/User-Manual/Audit-Log_timecho.md index 4909c77e8..cb2ff4cdd 100644 --- a/src/zh/UserGuide/V1.3.x/User-Manual/Audit-Log_timecho.md +++ b/src/zh/UserGuide/V1.3.x/User-Manual/Audit-Log_timecho.md @@ -20,7 +20,7 @@ --> -# 审计日志 +# 安全审计 ## 功能背景 diff --git a/src/zh/UserGuide/V1.3.x/User-Manual/Streaming_apache.md b/src/zh/UserGuide/V1.3.x/User-Manual/Streaming_apache.md index 839e036e3..14393c9f1 100644 --- a/src/zh/UserGuide/V1.3.x/User-Manual/Streaming_apache.md +++ b/src/zh/UserGuide/V1.3.x/User-Manual/Streaming_apache.md @@ -19,7 +19,7 @@ --> -# IoTDB 流处理框架 +# 流计算框架 IoTDB 流处理框架允许用户实现自定义的流处理逻辑，可以实现对存储引擎变更的监听和捕获、实现对变更数据的变形、实现对变形后数据的向外推送等逻辑。 diff --git a/src/zh/UserGuide/V1.3.x/User-Manual/Streaming_timecho.md b/src/zh/UserGuide/V1.3.x/User-Manual/Streaming_timecho.md index 64245aece..010ce7ce6 100644 --- a/src/zh/UserGuide/V1.3.x/User-Manual/Streaming_timecho.md +++ b/src/zh/UserGuide/V1.3.x/User-Manual/Streaming_timecho.md @@ -19,7 +19,7 @@ --> -# IoTDB 流处理框架 +# 流计算框架 IoTDB 流处理框架允许用户实现自定义的流处理逻辑，可以实现对存储引擎变更的监听和捕获、实现对变更数据的变形、实现对变形后数据的向外推送等逻辑。 diff --git a/src/zh/UserGuide/V1.3.x/User-Manual/User-defined-function_apache.md b/src/zh/UserGuide/V1.3.x/User-Manual/User-defined-function_apache.md index ad4c13f6c..e0816c3a9 100644 --- a/src/zh/UserGuide/V1.3.x/User-Manual/User-defined-function_apache.md +++ b/src/zh/UserGuide/V1.3.x/User-Manual/User-defined-function_apache.md @@ -1,4 +1,4 @@ -# 用户自定义函数 +# UDF ## 1. UDF 介绍 diff --git a/src/zh/UserGuide/V1.3.x/User-Manual/User-defined-function_timecho.md b/src/zh/UserGuide/V1.3.x/User-Manual/User-defined-function_timecho.md index ea47440fd..bacdb6fc4 100644 --- a/src/zh/UserGuide/V1.3.x/User-Manual/User-defined-function_timecho.md +++ b/src/zh/UserGuide/V1.3.x/User-Manual/User-defined-function_timecho.md @@ -1,4 +1,4 @@ -# 用户自定义函数 +# UDF ## 1. UDF 介绍 diff --git a/src/zh/UserGuide/dev-1.3/User-Manual/AINode_apache.md b/src/zh/UserGuide/dev-1.3/AI-capability/AINode_apache.md similarity index 99% rename from src/zh/UserGuide/dev-1.3/User-Manual/AINode_apache.md rename to src/zh/UserGuide/dev-1.3/AI-capability/AINode_apache.md index 6ff0a959f..b6620803f 100644 --- a/src/zh/UserGuide/dev-1.3/User-Manual/AINode_apache.md +++ b/src/zh/UserGuide/dev-1.3/AI-capability/AINode_apache.md @@ -19,7 +19,7 @@ --> -# AI能力（AINode） +# AINode AINode 是 IoTDB 在ConfigNode、DataNode后提供的第三种内生节点，该节点通过与 IoTDB 集群的 DataNode、ConfigNode 的交互，扩展了对时间序列进行机器学习分析的能力，支持从外部引入已有机器学习模型进行注册，并使用注册的模型在指定时序数据上通过简单 SQL 语句完成时序分析任务的过程，将模型的创建、管理及推理融合在数据库引擎中。目前已提供常见时序分析场景（例如预测与异常检测）的机器学习算法或自研模型。 @@ -33,7 +33,7 @@ AINode 是 IoTDB 在ConfigNode、DataNode后提供的第三种内生节点，该 - **DataNode**：负责接收并解析用户的 SQL请求；负责存储时间序列数据；负责数据的预处理计算。 - **AINode**：负责模型文件的导入创建以及模型推理。 -## 优势特点 +## 1. 优势特点 与单独构建机器学习服务相比，具有以下优势： @@ -49,7 +49,7 @@ AINode 是 IoTDB 在ConfigNode、DataNode后提供的第三种内生节点，该 - **时间序列标注（Time Series Annotation）**：为每个数据点或特定时间段添加额外的信息或标记，例如事件发生、异常点、趋势变化等，以便更好地理解和分析数据。 -## 基本概念 +## 2. 基本概念 - **模型（Model）**：机器学习模型，以时序数据作为输入，输出分析任务的结果或决策。模型是AINode 的基本管理单元，支持模型的增（注册）、删、查、用（推理）。 - **创建（Create）**: 将外部设计或训练好的模型文件或算法加载到MLNode中，由IoTDB统一管理与使用。 @@ -60,15 +60,15 @@ AINode 是 IoTDB 在ConfigNode、DataNode后提供的第三种内生节点，该 ::: -## 安装部署 +## 3. 安装部署 AINode 的部署可参考文档 [部署指导](../Deployment-and-Maintenance/AINode_Deployment_apache.md#ainode-部署) 章节。 -## 使用指导 +## 4. 使用指导 AINode 对时序数据相关的深度学习模型提供了模型创建及删除的流程，内置模型无需创建及删除，可直接使用，并且在完成推理后创建的内置模型实例将自动销毁。 -### 注册模型 +### 4.1 注册模型 通过指定模型输入输出的向量维度，可以注册训练好的深度学习模型，从而用于模型推理。 @@ -154,7 +154,7 @@ SQL执行后会异步进行注册的流程，可以通过模型展示查看模 模型注册完成后，就可以通过使用正常查询的方式调用具体函数，进行模型推理。 -### 查看模型 +### 4.2 查看模型 注册成功的模型可以通过show models指令查询模型的具体信息。其SQL定义如下： @@ -202,7 +202,7 @@ IoTDB> show models 我们前面已经注册了对应的模型，可以通过对应的指定查看模型状态，active表明模型注册成功，可用于推理。 -### 删除模型 +### 4.3 删除模型 对于注册成功的模型，用户可以通过SQL进行删除。该操作除了删除configNode上的元信息外，还会删除所有AINode下的相关模型文件。其SQL如下： @@ -212,7 +212,7 @@ drop model 需要指定已经成功注册的模型model_name来删除对应的模型。由于模型删除涉及多个节点上的数据删除，操作不会立即完成，此时模型的状态为DROPPING，该状态的模型不能用于模型推理。 -### 使用内置模型推理 +### 4.4 使用内置模型推理 SQL语法如下： @@ -281,7 +281,7 @@ IoTDB> call inference(_Stray, "select s0 from root.eg.airline", k=2) Total line number = 144 ``` -### 使用深度学习模型推理 +### 4.5 使用深度学习模型推理 SQL语法如下： @@ -441,7 +441,7 @@ Total line number = 4 其中结果集中每行的标签对应每24行数据为一组，输入该异常检测模型后的输出。 -## 权限管理 +## 5. 权限管理 使用AINode相关的功能时，可以使用IoTDB本身的鉴权去做一个权限管理，用户只有在具备 USE_MODEL 权限时，才可以使用模型管理的相关功能。当使用推理功能时，用户需要有访问输入模型的SQL对应的源序列的权限。 @@ -450,9 +450,9 @@ Total line number = 4 | USE_MODEL | create model / show models / drop model | √ | √ | x | | READ_DATA | call inference | √ | √ | √ | -## 实际案例 +## 6. 实际案例 -### 电力负载预测 +### 6.1 电力负载预测 在部分工业场景下，会存在预测电力负载的需求，预测结果可用于优化电力供应、节约能源和资源、支持规划和扩展以及增强电力系统的可靠性。 @@ -523,7 +523,7 @@ Total line number = 48 可以看到，我们使用了过去96个小时（4天）的六个负载信息和对应时间油温的关系，基于之前学习到的序列间相互关系对未来48个小时（2天）的油温这一数据的可能变化进行了建模，可以看到可视化后预测曲线与实际结果在趋势上保持了较高程度的一致性。 -### 功率预测 +### 6.2 功率预测 变电站需要对电流、电压、功率等数据进行电力监控，用于检测潜在的电网问题、识别电力系统中的故障、有效管理电网负载以及分析电力系统的性能和趋势等。 @@ -588,7 +588,7 @@ Total line number = 48 可以看到，我们使用了过去10分钟的电压的数据，基于之前学习到的序列间相互关系对未来5分钟的C相电压这一数据的可能变化进行了建模，可以看到可视化后预测曲线与实际结果在趋势上保持了一定的同步性。 -### 异常检测 +### 6.3 异常检测 在民航交通运输业，存在着对乘机旅客数量进行异常检测的需求。异常检测的结果可用于指导调整航班的调度，以使得企业获得更大效益。 diff --git a/src/zh/UserGuide/dev-1.3/User-Manual/AINode_timecho.md b/src/zh/UserGuide/dev-1.3/AI-capability/AINode_timecho.md similarity index 98% rename from src/zh/UserGuide/dev-1.3/User-Manual/AINode_timecho.md rename to src/zh/UserGuide/dev-1.3/AI-capability/AINode_timecho.md index 594236294..c67c41de5 100644 --- a/src/zh/UserGuide/dev-1.3/User-Manual/AINode_timecho.md +++ b/src/zh/UserGuide/dev-1.3/AI-capability/AINode_timecho.md @@ -19,7 +19,7 @@ --> -# AI能力（AINode） +# AINode AINode 是 IoTDB 在ConfigNode、DataNode后提供的第三种内生节点，该节点通过与 IoTDB 集群的 DataNode、ConfigNode 的交互，扩展了对时间序列进行机器学习分析的能力，支持从外部引入已有机器学习模型进行注册，并使用注册的模型在指定时序数据上通过简单 SQL 语句完成时序分析任务的过程，将模型的创建、管理及推理融合在数据库引擎中。目前已提供常见时序分析场景（例如预测与异常检测）的机器学习算法或自研模型。 @@ -33,7 +33,7 @@ AINode 是 IoTDB 在ConfigNode、DataNode后提供的第三种内生节点，该 - **DataNode**：负责接收并解析用户的 SQL请求；负责存储时间序列数据；负责数据的预处理计算。 - **AINode**：负责模型文件的导入创建以及模型推理。 -## 优势特点 +## 1. 优势特点 与单独构建机器学习服务相比，具有以下优势： @@ -49,7 +49,7 @@ AINode 是 IoTDB 在ConfigNode、DataNode后提供的第三种内生节点，该 - **时间序列标注（Time Series Annotation）**：为每个数据点或特定时间段添加额外的信息或标记，例如事件发生、异常点、趋势变化等，以便更好地理解和分析数据。 -## 基本概念 +## 2. 基本概念 - **模型（Model）**：机器学习模型，以时序数据作为输入，输出分析任务的结果或决策。模型是AINode 的基本管理单元，支持模型的增（注册）、删、查、用（推理）。 - **创建（Create）**: 将外部设计或训练好的模型文件或算法加载到MLNode中，由IoTDB统一管理与使用。 @@ -60,15 +60,15 @@ AINode 是 IoTDB 在ConfigNode、DataNode后提供的第三种内生节点，该 ::: -## 安装部署 +## 3. 安装部署 AINode 的部署可参考文档 [部署指导](../Deployment-and-Maintenance/AINode_Deployment_timecho.md#AINode-部署) 章节。 -## 使用指导 +## 4. 使用指导 AINode 对时序数据相关的深度学习模型提供了模型创建及删除的流程，内置模型无需创建及删除，可直接使用，并且在完成推理后创建的内置模型实例将自动销毁。 -### 注册模型 +### 4.1 注册模型 通过指定模型输入输出的向量维度，可以注册训练好的深度学习模型，从而用于模型推理。 @@ -154,7 +154,7 @@ SQL执行后会异步进行注册的流程，可以通过模型展示查看模 模型注册完成后，就可以通过使用正常查询的方式调用具体函数，进行模型推理。 -### 查看模型 +### 4.2 查看模型 注册成功的模型可以通过show models指令查询模型的具体信息。其SQL定义如下： @@ -202,7 +202,7 @@ IoTDB> show models 我们前面已经注册了对应的模型，可以通过对应的指定查看模型状态，active表明模型注册成功，可用于推理。 -### 删除模型 +### 4.3 删除模型 对于注册成功的模型，用户可以通过SQL进行删除。该操作除了删除configNode上的元信息外，还会删除所有AINode下的相关模型文件。其SQL如下： @@ -212,7 +212,7 @@ drop model 需要指定已经成功注册的模型model_name来删除对应的模型。由于模型删除涉及多个节点上的数据删除，操作不会立即完成，此时模型的状态为DROPPING，该状态的模型不能用于模型推理。 -### 使用内置模型推理 +### 4.4 使用内置模型推理 SQL语法如下： @@ -281,7 +281,7 @@ IoTDB> call inference(_Stray, "select s0 from root.eg.airline", k=2) Total line number = 144 ``` -### 使用深度学习模型推理 +### 4.5 使用深度学习模型推理 SQL语法如下： @@ -441,7 +441,12 @@ Total line number = 4 其中结果集中每行的标签对应每24行数据为一组，输入该异常检测模型后的输出。 -## 权限管理 + +### 4.6 时序大模型导入步骤 + +AINode 目前支持多种时序大模型，部署使用请参考[时序大模型](../AI-capability/TimeSeries-Large-Model) + +## 5. 权限管理 使用AINode相关的功能时，可以使用IoTDB本身的鉴权去做一个权限管理，用户只有在具备 USE_MODEL 权限时，才可以使用模型管理的相关功能。当使用推理功能时，用户需要有访问输入模型的SQL对应的源序列的权限。 @@ -450,9 +455,9 @@ Total line number = 4 | USE_MODEL | create model / show models / drop model | √ | √ | x | | READ_DATA | call inference | √ | √ | √ | -## 实际案例 +## 6. 实际案例 -### 电力负载预测 +### 6.1 电力负载预测 在部分工业场景下，会存在预测电力负载的需求，预测结果可用于优化电力供应、节约能源和资源、支持规划和扩展以及增强电力系统的可靠性。 @@ -523,7 +528,7 @@ Total line number = 48 可以看到，我们使用了过去96个小时（4天）的六个负载信息和对应时间油温的关系，基于之前学习到的序列间相互关系对未来48个小时（2天）的油温这一数据的可能变化进行了建模，可以看到可视化后预测曲线与实际结果在趋势上保持了较高程度的一致性。 -### 功率预测 +### 6.2 功率预测 变电站需要对电流、电压、功率等数据进行电力监控，用于检测潜在的电网问题、识别电力系统中的故障、有效管理电网负载以及分析电力系统的性能和趋势等。 @@ -588,7 +593,7 @@ Total line number = 48 可以看到，我们使用了过去10分钟的电压的数据，基于之前学习到的序列间相互关系对未来5分钟的C相电压这一数据的可能变化进行了建模，可以看到可视化后预测曲线与实际结果在趋势上保持了一定的同步性。 -### 异常检测 +### 6.3 异常检测 在民航交通运输业，存在着对乘机旅客数量进行异常检测的需求。异常检测的结果可用于指导调整航班的调度，以使得企业获得更大效益。 diff --git a/src/zh/UserGuide/dev-1.3/AI-capability/TimeSeries-Large-Model.md b/src/zh/UserGuide/dev-1.3/AI-capability/TimeSeries-Large-Model.md new file mode 100644 index 000000000..cf7804394 --- /dev/null +++ b/src/zh/UserGuide/dev-1.3/AI-capability/TimeSeries-Large-Model.md @@ -0,0 +1,111 @@ + + +# 时序大模型 + +## 简介 + +时序大模型是一种专为时序数据分析设计的基础模型。IoTDB 团队长期自研时序大模型，基于变换器（Transformer）结构等技术在海量时序数据上预训练，能够理解并生成多种领域的时序数据，可被应用于时序预测、异常检测、时序填补等应用场景。不同于传统时序分析技术，时序大模型具备通用特征提取能力，基于零样本分析、微调等技术服务广泛的分析任务。 + +团队所研时序大模型相关技术均发表在国际机器学习顶级会议。 + +## 应用场景 + +- **时序预测**：为工业生产、自然环境等领域提供时间序列数据的预测服务，帮助用户提前了解未来趋势。 +- **数据填补**：针对时间序列中的缺失序列段，进行上下文填补，以增强数据集的连续性和完整性。 +- **异常检测**：利用自回归分析技术，对时间序列数据进行实时监测，及时预警潜在的异常情况。 + +![](/img/LargeModel09.png) + +## Timer 模型 + +Timer模型不仅展现了出色的少样本泛化和多任务适配能力，还通过预训练获得了丰富的知识库，赋予了它处理多样化下游任务的通用能力，拥有以下特点： + +- **泛化性**：模型能够通过使用少量样本进行微调，达到行业内领先的深度模型预测效果。 +- **通用性**：模型设计灵活，能够适配多种不同的任务需求，并且支持变化的输入和输出长度，使其在各种应用场景中都能发挥作用。 +- **可扩展性**：随着模型参数数量的增加或预训练数据规模的扩大，模型的性能会持续提升，确保模型能够随着时间和数据量的增长而不断优化其预测效果。 + +![](/img/LargeModel02.png) + +## Timer-XL 模型 + +Timer-XL 基于 Timer 进一步扩展升级了网络结构，在多个维度上进行全面突破： + +- **超长上下文支持**：该模型突破了传统时序预测模型的限制，支持处理数千个Token（相当于数万个时间点）的输入，有效解决了上下文长度的瓶颈问题。 +- **多变量预测场景覆盖**：支持多种预测场景，包括非平稳时间序列的预测、涉及多个变量的预测任务以及包含协变量的预测，满足多样化的业务需求。 +- **大规模工业时序数据集：**采用万亿大规模工业物联网领域的时序数据集进行预训练，数据集兼有庞大的体量、卓越的质量和丰富的领域等重要特质，覆盖能源、航空航天、钢铁、交通等多领域。 + +## 效果展示 + +时序大模型能够适应多种不同领域和场景的真实时序数据，在各种任务上拥有优异的处理效果，以下是在不同数据上的真实表现： + +**时序预测：** + +利用时序大模型的预测能力，能够准确预测时间序列的未来变化趋势，如下图蓝色曲线代表预测趋势，红色曲线为实际趋势，两曲线高度吻合。 + +![](/img/LargeModel03.png) + +**数据填补：**： + +利用时序大模型对缺失数据段进行预测式填补。 + +![](/img/LargeModel04.png) + + +**异常检测：**： + +利用时序大模型精准识别与正常趋势偏离过大的异常值。 + +![](/img/LargeModel05.png) + +## 部署使用 + +1. 打开 IoTDB cli 控制台，检查 ConfigNode、DataNode、AINode 节点确保均为 Running。 + +检查命令： +```sql +show cluster +``` + +![](/img/ainode-timer-1.png) + +2. 模型文件存放路径：推荐放在 AINode 安装包相同文件夹下，可新建模型文件夹存放模型文件 +3. 注册模型语句 + +```sql +create model using uri +``` + +示例： + +```sql +create model Timer-xl using uri +``` + +4. 检查模型是否注册成功 + +检查命令： + +```sql +show models +``` + +![](/img/LargeModel06.png) diff --git a/src/zh/UserGuide/dev-1.3/User-Manual/Audit-Log_timecho.md b/src/zh/UserGuide/dev-1.3/User-Manual/Audit-Log_timecho.md index 4909c77e8..cb2ff4cdd 100644 --- a/src/zh/UserGuide/dev-1.3/User-Manual/Audit-Log_timecho.md +++ b/src/zh/UserGuide/dev-1.3/User-Manual/Audit-Log_timecho.md @@ -20,7 +20,7 @@ --> -# 审计日志 +# 安全审计 ## 功能背景 diff --git a/src/zh/UserGuide/dev-1.3/User-Manual/Streaming_apache.md b/src/zh/UserGuide/dev-1.3/User-Manual/Streaming_apache.md index 839e036e3..14393c9f1 100644 --- a/src/zh/UserGuide/dev-1.3/User-Manual/Streaming_apache.md +++ b/src/zh/UserGuide/dev-1.3/User-Manual/Streaming_apache.md @@ -19,7 +19,7 @@ --> -# IoTDB 流处理框架 +# 流计算框架 IoTDB 流处理框架允许用户实现自定义的流处理逻辑，可以实现对存储引擎变更的监听和捕获、实现对变更数据的变形、实现对变形后数据的向外推送等逻辑。 diff --git a/src/zh/UserGuide/dev-1.3/User-Manual/Streaming_timecho.md b/src/zh/UserGuide/dev-1.3/User-Manual/Streaming_timecho.md index 64245aece..010ce7ce6 100644 --- a/src/zh/UserGuide/dev-1.3/User-Manual/Streaming_timecho.md +++ b/src/zh/UserGuide/dev-1.3/User-Manual/Streaming_timecho.md @@ -19,7 +19,7 @@ --> -# IoTDB 流处理框架 +# 流计算框架 IoTDB 流处理框架允许用户实现自定义的流处理逻辑，可以实现对存储引擎变更的监听和捕获、实现对变更数据的变形、实现对变形后数据的向外推送等逻辑。 diff --git a/src/zh/UserGuide/dev-1.3/User-Manual/User-defined-function_apache.md b/src/zh/UserGuide/dev-1.3/User-Manual/User-defined-function_apache.md index cb034eb4e..f95b12e38 100644 --- a/src/zh/UserGuide/dev-1.3/User-Manual/User-defined-function_apache.md +++ b/src/zh/UserGuide/dev-1.3/User-Manual/User-defined-function_apache.md @@ -1,4 +1,4 @@ -# 用户自定义函数 +# UDF ## 1. UDF 介绍 diff --git a/src/zh/UserGuide/dev-1.3/User-Manual/User-defined-function_timecho.md b/src/zh/UserGuide/dev-1.3/User-Manual/User-defined-function_timecho.md index b53014e15..e4add1fde 100644 --- a/src/zh/UserGuide/dev-1.3/User-Manual/User-defined-function_timecho.md +++ b/src/zh/UserGuide/dev-1.3/User-Manual/User-defined-function_timecho.md @@ -1,4 +1,4 @@ -# 用户自定义函数 +# UDF ## 1. UDF 介绍 diff --git a/src/zh/UserGuide/latest/User-Manual/Audit-Log_timecho.md b/src/zh/UserGuide/latest/User-Manual/Audit-Log_timecho.md index 4f88f5732..3fd2eac0d 100644 --- a/src/zh/UserGuide/latest/User-Manual/Audit-Log_timecho.md +++ b/src/zh/UserGuide/latest/User-Manual/Audit-Log_timecho.md @@ -20,7 +20,7 @@ --> -# 审计日志 +# 安全审计 ## 1. 功能背景 diff --git a/src/zh/UserGuide/latest/User-Manual/Streaming_apache.md b/src/zh/UserGuide/latest/User-Manual/Streaming_apache.md index 8e7fd3e58..fa974851d 100644 --- a/src/zh/UserGuide/latest/User-Manual/Streaming_apache.md +++ b/src/zh/UserGuide/latest/User-Manual/Streaming_apache.md @@ -19,7 +19,7 @@ --> -# 流处理框架 +# 流计算框架 IoTDB 流处理框架允许用户实现自定义的流处理逻辑，可以实现对存储引擎变更的监听和捕获、实现对变更数据的变形、实现对变形后数据的向外推送等逻辑。 diff --git a/src/zh/UserGuide/latest/User-Manual/Streaming_timecho.md b/src/zh/UserGuide/latest/User-Manual/Streaming_timecho.md index b8ecd936e..a71bc7361 100644 --- a/src/zh/UserGuide/latest/User-Manual/Streaming_timecho.md +++ b/src/zh/UserGuide/latest/User-Manual/Streaming_timecho.md @@ -19,7 +19,7 @@ --> -# IoTDB 流处理框架 +# 流计算框架 IoTDB 流处理框架允许用户实现自定义的流处理逻辑，可以实现对存储引擎变更的监听和捕获、实现对变更数据的变形、实现对变形后数据的向外推送等逻辑。 diff --git a/src/zh/UserGuide/latest/User-Manual/User-defined-function_apache.md b/src/zh/UserGuide/latest/User-Manual/User-defined-function_apache.md index 375b089f6..6289c6808 100644 --- a/src/zh/UserGuide/latest/User-Manual/User-defined-function_apache.md +++ b/src/zh/UserGuide/latest/User-Manual/User-defined-function_apache.md @@ -1,4 +1,4 @@ -# 用户自定义函数 +# UDF ## 1. UDF 介绍 diff --git a/src/zh/UserGuide/latest/User-Manual/User-defined-function_timecho.md b/src/zh/UserGuide/latest/User-Manual/User-defined-function_timecho.md index de1401d3b..1c414e80a 100644 --- a/src/zh/UserGuide/latest/User-Manual/User-defined-function_timecho.md +++ b/src/zh/UserGuide/latest/User-Manual/User-defined-function_timecho.md @@ -1,4 +1,4 @@ -# 用户自定义函数 +# UDF ## 1. UDF 介绍