A multi-condition time series data cycle analysis method and system based on AI technology

Abstract

The application provides a multi-condition time series data cycle analysis method and system based on AI technology. Through modular engineering design, multiple links such as multi-condition combination generation, data preprocessing, AI intelligent analysis, task scheduling and result processing are integrated to form a configurable and automated complete analysis method. The system is composed of a configuration module, a task module, an AI analysis module, a control module, a storage module and a tool package. The system can process a large amount of condition combination data in parallel and automatically identify the cycle characteristics of time series data using an AI model, and finally generate a structured analysis result. The beneficial effects of the application are that the analysis efficiency of time series data can be effectively improved, more comprehensive analysis results can be provided, and the use threshold can be reduced.

Description

Technical Field

[0001] This invention belongs to the field of artificial intelligence technology, and in particular relates to a method and system for multi-condition time series data periodic analysis based on AI technology. Background Technology

[0002] Traditional time series data analysis methods typically rely on a series of manual operations by analysts. First, the time interval and combination of conditions to be analyzed are determined based on empirical criteria. Then, data processing is performed, including time window division and time series value calculation. Based on this, periodic analysis is conducted using statistical methods or empirical judgment to identify potential periodic patterns in the data. Finally, the identified periods are manually analyzed and summarized. This method is not only cumbersome and time-consuming, but also limited by manual processing capabilities, often failing to cover multiple combinations of conditions. It usually only allows for sampling analysis, limiting the comprehensiveness of the results and placing a high reliance on the professional experience of the operators.

[0003] In summary, the technical problem to be solved is:

[0004] 1. Overcome the inefficiency of traditional time series data analysis methods.

[0005] 2. To address the issue that sampling analysis methods cannot provide comprehensive coverage.

[0006] 3. It addresses the problem that traditional analytical methods rely on experience and professional knowledge, resulting in high barriers to entry. Summary of the Invention

[0007] The purpose of this invention is to address the shortcomings of existing technologies by providing a method and system for multi-condition time series data periodic analysis based on AI technology.

[0008] The objective of this invention is achieved through the following technical solution: a multi-condition time-series data periodic analysis system based on AI technology, comprising the following modules:

[0009] The configuration module is used to read the global configuration file provided by the user and generate configuration information objects for use by the control module.

[0010] The AI ​​analysis module loads a periodic analysis model, calculates the periodic significance index based on the time series data sequence, identifies one or more periods in the data, and submits the analysis results to the task module.

[0011] The task module takes the configuration and data of a specific task as input, works in conjunction with the AI ​​analysis module, and outputs the periodic characteristics of the data under corresponding conditions; the data includes condition columns, time series, and numerical columns.

[0012] The control module calls the configuration module to complete initialization after the system starts. It decomposes and schedules tasks based on the combination of each condition in the condition column, and organizes and collects the execution results.

[0013] The storage module is responsible for the persistence of file-level tasks and log data.

[0014] Furthermore, the global configuration file includes input / output configuration, condition configuration, preprocessing configuration, and analysis configuration. The input / output configuration specifies the data source file path for system input and the template portion of the output path. In a multi-task scenario, the output path will be divided into multiple directories according to the template and conditions. The condition configuration is used to set the combination of conditions and filtering settings for the data filtering process. The preprocessing configuration sets the grouping and aggregation settings for the filtered data, including the effective range, group value range, and numerical columns. The analysis configuration sets the loading settings for the periodic analysis model, including the model type, model path, and model parameters.

[0015] Furthermore, the periodic analysis model is a time series analysis model based on the Transformer architecture.

[0016] Furthermore, the task module includes a data processing module, a task information integration module, and a task execution module;

[0017] Data processing module: Aggregates data in units of time windows to generate time series data sequences suitable for model input;

[0018] Task execution module: For each task, it calls the interfaces provided by the data processing module and the AI ​​analysis module to obtain the analysis results, and uses the task information integration module to encapsulate the information related to the output throughout the process. Finally, it ends the task execution and returns the encapsulated results.

[0019] Task information integration module: During task execution, this module includes task configuration information, task execution process information, and task results. It binds and encapsulates this information into task result information.

[0020] Furthermore, the control module includes the following components:

[0021] System initialization: After the system starts, it enters the control module, which first calls the configuration module to obtain the system configuration object; then, it obtains the input data information based on the configuration object and reads it into the system as a data object; subsequently, it allocates the information in the configuration object to assist the task module, AI analysis module, and storage module in completing their respective initialization work; the initialization of the control module itself is also carried out simultaneously, including task thread resource pooling, task management initialization, and log function initialization.

[0022] Task scheduling: First, the task is decomposed. The control module generates specific filtering conditions based on the configuration and divides the overall data into corresponding task subsets. A globally unique ID is generated for each task, and a task execution object is created. The storage module is called to generate a separate storage space for the task. Then, a multi-threaded task execution process is initiated, and the thread pool and task module are scheduled to perform analysis jobs. During task execution, this module is responsible for monitoring the running status of the initiated tasks and maintaining the running information.

[0023] Task Management: A task list is maintained in memory, recording the initial, running, and result information of each initiated task. The lifecycle information of each task, including task ID, condition settings, execution results, and exception information, is synchronized to the task list. After a single task is completed, this function synchronizes the corresponding information in the task list to the storage area. When all tasks are completed, this function organizes and aggregates all entries in the list to generate a unified execution report.

[0024] Furthermore, the storage module includes an analysis storage area, a task storage area, and a log storage area; the task storage area establishes storage blocks indexed by task IDs, and each block stores the data conditions, intermediate results, task termination status, final analysis results, and error information for a single task. The system's operation report is also stored in the task storage area; the log storage area contains the system operation logs, and each log file is a summary of a single system execution, including the overall system running time, the number of tasks executed, and the overall execution status.

[0025] The toolkit includes a results aggregation tool that extracts and summarizes the results of all tasks executed in the same session from the task storage area, and aggregates them into the analysis storage area of ​​the storage module according to conditions to form a structured database table.

[0026] Furthermore, the toolkit includes a condition combination tool, an auxiliary generation tool for configuring conditions, used to combine condition column names of input data.

[0027] This invention also provides a method for multi-condition time series data periodic analysis based on AI technology, implemented using the aforementioned AI-based multi-condition time series data periodic analysis system, comprising the following steps:

[0028] Create a global configuration file that contains input / output configuration, condition configuration, preprocessing configuration, and analysis configuration;

[0029] During system initialization, the control module calls the task module, AI analysis module, and storage module to complete their respective initialization tasks; the control module itself is also initialized simultaneously, including task thread resource pooling, task management initialization, and log function initialization.

[0030] The control module reads the condition configuration information in the configuration, splits it into multiple tasks according to the configuration, and divides and filters the dataset according to the filtering conditions; it synchronously creates a task list, generates a unique ID for each combination of conditions, and creates a task execution object and an independent directory for each task; for each task, it allocates thread resources to the task in batches and executes them in the order of data processing, AI analysis, and task information integration.

[0031] The task management function of the control module receives the result object returned by each task, updates the task list and writes it to the dedicated task storage area directory; when all tasks are completed, a unified execution report is generated and stored in the task storage area.

[0032] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the above-described method for multi-condition time-series data periodic analysis based on AI technology.

[0033] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described AI-based multi-condition time-series data periodic analysis method.

[0034] Compared with the prior art, the system described in this invention has the following significant advantages:

[0035] 1. High efficiency and automation: Through engineering and modular system design, scattered analysis steps are integrated into automated processes, which greatly reduces manual operation and improves analysis efficiency.

[0036] 2. Broad analytical coverage: By combining multiple conditions and using concurrent techniques, it achieves coverage of a large number of analytical conditions, avoiding the limitations of traditional methods that can only perform sampling analysis due to excessive workload, and ensuring the comprehensiveness of the analytical results.

[0037] 3. Intelligent and experience-free: The AI ​​model is used to automatically detect the cycle, avoiding reliance on users' professional knowledge and experience in time series analysis, thus lowering the barrier to entry.

[0038] 4. Results are manageable and analyzable: Through structured storage and a dedicated post-processing module, the system can effectively manage and collect large amounts of analytical results, facilitating subsequent research. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 A software architecture diagram of a multi-condition time-series data periodic analysis system based on AI technology is provided for embodiments of the present invention.

[0041] Figure 2 A data flow diagram of a multi-condition time-series data periodic analysis system based on AI technology provided in an embodiment of the present invention;

[0042] Figure 3 A flowchart of a multi-condition time series data periodic analysis method based on AI technology is provided for an embodiment of the present invention. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the scope of protection of the invention. It should be noted that, unless otherwise specified, the following embodiments and features can be combined with each other.

[0044] This invention provides a method and system for multi-condition time-series data periodic analysis based on AI technology. For example... Figure 1 , Figure 2 and Figure 3 As shown, through modular engineering design, multiple stages such as multi-condition combination generation, data preprocessing, AI intelligent analysis, task scheduling, and result processing are integrated to form a configurable and automated complete analysis process. This invention can process large amounts of condition combination data in parallel, automatically identify the periodic characteristics of time-series data using AI models, and ultimately generate structured analysis results, thereby effectively improving the efficiency and coverage of analysis and lowering the barrier to entry.

[0045] Figure 1 A schematic diagram of a multi-condition time-series data periodic analysis system based on AI technology, provided for an embodiment of the present invention, includes:

[0046] 1. Configuration Module

[0047] The configuration module reads the user-provided global configuration file. This configuration file uses a template design and defines the basic parameters of the analysis task. This module is responsible for parsing the configuration information and generating configuration information objects that can be used by the control module based on the template.

[0048] The basic configuration includes input / output configuration, condition configuration, preprocessing configuration, and analysis configuration. Input / output configuration primarily specifies the data source file path for system input and the template for the output path. In multi-task scenarios, the output path will be divided into multiple directories based on this template and the conditions. Condition configuration mainly sets the combination of conditions and filtering settings for the data filtering stage. Preprocessing configuration mainly sets the grouping and aggregation of the filtered data, including the effective range, group value range, and numerical columns. Analysis configuration mainly sets the loading of the analysis model, including the model type, model path, and model parameters.

[0049] This module reads configuration files to create configuration information objects in the system, providing setting support for system operation.

[0050] 2. Task Module

[0051] The task module is the core of a single task execution function. Its inputs are the specific task's configuration and data, and its output is the periodic characteristics of the data under specific conditions. This module needs to work in conjunction with the AI ​​analysis module, and its main functions include data processing, task execution, and runtime information management.

[0052] 2.1. Data Processing

[0053] The main function of the data processing module is to perform preprocessing and transformation operations on the data according to the configuration settings. First, it divides the data into time windows according to the settings to control the granularity of the data analysis. Then, it aggregates the data using time windows as units (such as summation, averaging, counting, etc.) to generate a well-organized time-series data sequence suitable for model input.

[0054] 2.2. Task Information Integration

[0055] The task execution process includes task configuration information, task execution process information, and task results. The role of task information integration is to bind this information and encapsulate it into task result information according to the module output standards, thus maintaining the data link of the overall process.

[0056] 2.3. Task Execution

[0057] This function is the main control unit for a single task, responsible for task initialization, execution, and result delivery. It first prepares the task based on the task configuration information (which is a subset of input / output, condition, preprocessing, and analysis configurations, read from the configuration module). This includes setting initialization and variable initialization. Then, it executes the task, obtaining analysis results by calling interfaces provided by the data processing and AI analysis modules. Throughout the process, it uses task information integration to encapsulate output-related information. Finally, it terminates task execution and returns the encapsulated results.

[0058] 3. AI Analysis Module

[0059] This module can load trained periodic analysis models (such as time series analysis models based on the Transformer architecture). It can calculate the periodic significance index of time series data sequences, identify one or more major periods in the data, and submit the analysis results to the upstream caller.

[0060] 3.1. Model Loading

[0061] This feature supports loading time series analysis models according to model category, model path, and optional configuration information. Loaded models will mask category differences, providing a unified calling method. The loading interface provides default parameters for each model category, ensuring both minimal configuration operation and custom deployment with adjustable parameters. Adhering to the open / closed design principle, the model loading function has a standard design interface, supporting loading extensions for other models.

[0062] 3.2. Analysis Interface

[0063] This interface serves as the call interface for the AI ​​analysis module, featuring standardized input and output definitions. This design aims to shield the differences in usage between different types of models, ensuring functional consistency throughout the program flow.

[0064] 4. Control Module

[0065] As the system's central hub, this module is responsible for overall task scheduling and system lifecycle management. After system startup, it calls the configuration module to complete initialization, performs task decomposition and execution scheduling, and uniformly organizes and aggregates execution results. In addition, this module is also responsible for log system management and operational status monitoring.

[0066] 4.1 System Initialization

[0067] This system is a batch processing system. The initialization process mainly involves configuration loading and input data loading. After the system starts, it enters the control module, which first calls the configuration module to obtain the system configuration object. Then, it retrieves the input data information based on the configuration object and reads it into the system as a data object.

[0068] Subsequently, this module allocates the information in the configuration object, assisting the task module, AI analysis module, and storage module in completing their respective initialization tasks. The control module itself is also initialized simultaneously, including task thread resource pooling, task management initialization, and logging function initialization.

[0069] 4.2. Task Scheduling

[0070] First, the task is decomposed. The control module generates specific filtering conditions based on the configuration and divides the overall data into corresponding task subsets. A globally unique ID is generated for each task, and a task execution object is created. The storage module is then called to generate separate storage space for each task.

[0071] Then, a multi-threaded task execution process is initiated, scheduling the thread pool and task module to perform analysis jobs. During task execution, this module is responsible for monitoring the running status of the initiated tasks and maintaining running information.

[0072] 4.3. Task Management

[0073] The task management function maintains a task list in memory, recording the initial information, running information, and result information (including execution results and exception information) for each initiated task. The lifecycle information of each task, including task ID, condition settings, execution results, and exception information, is synchronized to the task list. After a single task completes, the function synchronizes the corresponding information in the task list to storage. When all tasks have completed, the function organizes and aggregates all entries in the list to generate a unified execution report.

[0074] 5. Storage module

[0075] In addition to handling the persistence of file-level task and log data, this module also provides a structured analysis area, which, with the assistance of the toolkit, supports the export and analysis of results data. The storage module is divided into a task storage area, a log storage area, and an analysis storage area.

[0076] The task storage area and log storage area are primarily file-level storage. The task storage area uses task IDs as indexes to create storage blocks. Each block stores the data conditions, intermediate results, task termination status, final analysis results, and error messages for a single task. System operation reports are also stored in the task storage area. The log storage area mainly contains system operation logs. Each log file is a summary of a single system execution, including overall system runtime, number of tasks executed, and overall execution status.

[0077] The analysis storage area primarily uses structured storage, providing support similar to relational databases. Although the system has already collected and organized the results, in practice, further analysis and exploratory experiments are often required. This area is established for this purpose, facilitating analysts to transform the results and perform related queries.

[0078] 6. Toolkit

[0079] The main function of the toolkit is to provide support for the preparatory work and subsequent exploration of the system by providing system-independent software tools, mainly including condition combination tools and result aggregation tools.

[0080] The condition combination tool is an auxiliary tool for generating condition configurations. Condition configuration often requires combining multiple conditions. When there are many condition columns, manually constructing these combinations is inefficient, while fully automated generation lacks flexibility and control. The condition combination tool allows you to combine the condition column names of the input data to generate a list of conditions required for system operation, improving convenience.

[0081] The results aggregation tool is responsible for integrating and processing large amounts of scattered analysis results. Its main function is to extract and summarize the results of all tasks executed in the same session from the task storage area, and generate data that supports structured queries by associating it with runtime conditions (i.e., the result of combining multiple conditions, calculated by the condition combination tool). Finally, a data table is built in the analysis storage area. Users can export this structured result or analyze it directly.

[0082] Example 1: Multi-condition periodic analysis of global meteorological data based on AI technology

[0083] The main content of this embodiment is to use this system to study the periodic patterns of temperature changes under different climate zones and topography around the world. By automatically and in parallel analyzing temperature sequences under combinations of climate zones and topography, the main cycle of each combination is quickly identified, and structured result data is generated.

[0084] 1. Data Preparation

[0085] The global weather station observation data table includes the following key columns:

[0086] Condition columns: climate_zone, topography

[0087] Time series: timestamp

[0088] Numerical column: temperature (daily average temperature)

[0089] 2. System Configuration

[0090] Create a global configuration file named config.ini, which contains input / output configurations, condition configurations, preprocessing configurations, and analysis configurations. Input / output configurations include the input file path and output root directory. Condition configurations mainly consist of a list of filtering conditions and filtering criteria. Preprocessing configurations include the data processing save path, aggregation window size (one month in this example), aggregation data save path, aggregation data filtering criteria, and target value column. Analysis configurations include the model type, model path, model parameters, and results save path.

[0091] When constructing the condition filter list, the condition combination tool in the toolkit was used. This analysis not only examines the combined effect of climate zones and topography, but also the individual effect of each condition. Therefore, two conditions were input into the condition combination tool to generate the condition list:

[0092] ['climate_zone', ' topography', 'climate_zone, topography']

[0093] When the system is running, it will further decompose the tasks according to the corresponding values ​​in the data for each condition in the list. For example, in this embodiment, there are 13 climate zones (tropical rainforest climate, tropical savanna climate, tropical desert climate, tropical monsoon climate, subtropical monsoon climate, Mediterranean climate, temperate monsoon climate, temperate oceanic climate, temperate continental climate, subarctic coniferous forest climate, and polar tundra climate), which correspond to 13 operating conditions. Therefore, the 'climate_zone' item in the list can be split into 13 tasks. There are 5 types of terrain (plains, hills, mountains, plateaus, and basins), which correspond to 5 operating conditions. Therefore, the 'topography' item in the list can be split into 5 tasks. The combination of climate zones and terrain corresponds to 65 operating conditions, so the 'climate_zone, topography' item in the list can be split into 65 tasks.

[0094] 3. System Execution Flow

[0095] Step 1: System Initialization

[0096] Upon system startup, the control module calls the configuration module, successfully parsing `config.ini` and generating a configuration object in memory. Based on the file path input in the configuration object, the control module loads all data from the global weather station observation data tables into memory, forming a data object. Subsequently, the module splits the configuration object, calling the task module, AI analysis module, and storage module to complete their respective initialization tasks. The control module's own initialization also occurs simultaneously, including task thread resource pooling, task management initialization, and logging function initialization.

[0097] Step 2: Task Scheduling and Execution

[0098] The control module reads the conditional configuration information from the configuration file, breaks down the task into combinations of all climate zones and terrain features, and partitions and filters the dataset according to the filtering criteria. At the start of this process, a task list is created synchronously, generating a unique ID (e.g., task_001, task_002, etc.) for each condition combination; a task execution object is created for each task, containing the task's configuration information and the corresponding data subset. The storage module is then invoked to create an independent directory for each task in the storage area.

[0099] Tasks are allocated thread resources in batches and executed in parallel by calling the task module. For each task, data preprocessing is performed first, aggregating and calculating the data based on the set time unit and daily average temperature to generate a time-average temperature sequence. Then, the AI ​​analysis module is called to analyze the sequence using the loaded mode_v3 model. The model automatically analyzes the sequence, identifies the period with the most significant energy, and calculates the evaluation score. Finally, the task information integration function binds this result with the task ID, condition combination, running status, etc., encapsulates it into a standardized task result object, and returns it to the control module.

[0100] Step 3: Results Collection and Report Generation

[0101] The task management function of the control module receives the result object returned by each task and updates the task list. When a task is completed, the control module writes all information of the task (configuration, results, and running information) to its dedicated task storage directory through the storage module.

[0102] Once all tasks have been completed, the control module extracts key information from all entries in the task list (task ID, condition combination, running status, main cycle, evaluation score, and error information) and generates a unified JSON-formatted execution report, which is then stored in the task storage area.

[0103] Step 5: Post-analysis support

[0104] Researchers used the results aggregation tool in the toolkit. This tool, combined with the execution report, automatically scanned all subdirectories under the output root directory, read the record information for each task, and aggregated all results according to conditions into the analysis storage area of ​​the storage module, forming a structured database table analysis_results.

[0105] Researchers can directly query this table using SQL to quickly gain insights, for example:

[0106] -- Query all climate zones with "mountainous" terrain and a main cycle longer than 12 months, and their corresponding cycle distributions:

[0107] SELECT climate_zone, topography, primary_periods FROM analysis_results

[0108] WHERE topography = 'Mountain' AND primary_periods>12.

[0109] In another embodiment, this application is also applicable to multi-condition periodic analysis of groundwater detection data, which can be performed by referring to the above execution process; wherein the groundwater detection data includes: time series: detection time; condition column: aquifer type, monitoring level, whether it is abnormal; numerical column: water level depth, water temperature, mineralization.

[0110] Those skilled in the art will understand that this application can analyze any data that has conditional and temporal characteristics, and will not be elaborated further here.

[0111] Based on the same concept, this invention also provides a method for multi-condition time series data periodic analysis based on AI technology, implemented using the aforementioned multi-condition time series data periodic analysis system based on AI technology, including the following steps:

[0112] Create a global configuration file that contains input / output configuration, condition configuration, preprocessing configuration, and analysis configuration;

[0113] During system initialization, the control module calls the task module, AI analysis module, and storage module to complete their respective initialization tasks; the control module itself is also initialized simultaneously, including task thread resource pooling, task management initialization, and log function initialization.

[0114] The control module reads the condition configuration information in the configuration, splits it into multiple tasks according to the configuration, and divides and filters the dataset according to the filtering conditions; it synchronously creates a task list, generates a unique ID for each combination of conditions, and creates a task execution object and an independent directory for each task; for each task, it allocates thread resources to the task in batches and executes them in the order of data processing, AI analysis, and task information integration.

[0115] The task management function of the control module receives the result object returned by each task, updates the task list and writes it to the dedicated task storage area directory; when all tasks are completed, a unified execution report is generated and stored in the task storage area.

[0116] It should be noted that the method embodiment shown in this embodiment matches the content of the above system embodiment, and the content of the above system embodiment can be referred to, and will not be repeated here.

[0117] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the above-described method for multi-condition time-series data periodic analysis based on AI technology.

[0118] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described AI-based multi-condition time-series data periodic analysis method.

[0119] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0120] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0121] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0122] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0123] The above embodiments are only used to illustrate the design concept and features of the present invention, and their purpose is to enable those skilled in the art to understand the content of the present invention and implement it accordingly. The protection scope of the present invention is not limited to the above embodiments. Therefore, all equivalent changes or modifications made based on the principles and design ideas disclosed in the present invention are within the protection scope of the present invention.

Claims

1. A multi-condition time-series data periodic analysis system based on AI technology, characterized in that, Includes the following modules: The configuration module reads the global configuration file provided by the user and generates configuration information objects for use by the control module. The global configuration file includes input / output configuration, condition configuration, preprocessing configuration, and analysis configuration. The input / output configuration specifies the data source file path for system input and the template part of the output path. In multi-task scenarios, the output path will be divided into multiple directories according to the template and conditions. The condition configuration sets the combination of conditions and filtering settings for the data filtering process. The preprocessing configuration sets the grouping and aggregation of the filtered data, including the effective range, group value range, and numerical columns. The analysis configuration sets the loading of the periodic analysis model, including the model type, model path, and model parameters. The AI ​​analysis module loads a periodic analysis model, calculates the periodic significance index based on the time series data sequence, identifies one or more periods in the data, and submits the analysis results to the task module. The task module takes the configuration and data of a specific task as input, works in conjunction with the AI ​​analysis module, and outputs the periodic characteristics of the data under corresponding conditions; the data includes condition columns, time series, and numerical columns. The control module calls the configuration module to complete initialization after the system starts. It decomposes and schedules tasks based on the combination of each condition in the condition column, and organizes and collects the execution results. The storage module is responsible for the persistence of file-level task and log data. The storage module includes an analysis storage area, a task storage area, and a log storage area. The task storage area uses task IDs as indexes to create storage blocks. Each block stores the data conditions, intermediate results, task termination status, final analysis results, and error information for a single task. The system's operation report is also stored in the task storage area. The log storage area contains system operation logs. Each log file is a summary of a single system execution, including the overall system runtime, the number of tasks executed, and the overall execution status. The toolkit includes a result aggregation tool that extracts and summarizes all task results from the same execution in the task storage area, and aggregates them into the analysis storage area of ​​the storage module according to conditions to form a structured database table. The toolkit also includes a condition combination tool, an auxiliary generation tool for configuring conditions, used to combine condition column names of input data.

2. The multi-condition time-series data periodic analysis system based on AI technology according to claim 1, characterized in that, The periodic analysis model is a time series analysis model based on the Transformer architecture.

3. The multi-condition time-series data periodic analysis system based on AI technology according to claim 1, characterized in that, The task module includes a data processing module, a task information integration module, and a task execution module; Data processing module: Aggregates data in units of time windows to generate time series data sequences suitable for model input; Task execution module: For each task, it calls the interfaces provided by the data processing module and the AI ​​analysis module to obtain the analysis results, and uses the task information integration module to encapsulate the information related to the output throughout the process. Finally, it ends the task execution and returns the encapsulated results. Task information integration module: During task execution, this module includes task configuration information, task execution process information, and task results. It binds and encapsulates this information into task result information.

4. The multi-condition time-series data periodic analysis system based on AI technology according to claim 1, characterized in that, The control module includes the following components: System initialization: After the system starts, it enters the control module, which first calls the configuration module to obtain the system configuration object; then, it obtains the input data information based on the configuration object and reads it into the system as a data object; subsequently, it allocates the information in the configuration object to assist the task module, AI analysis module, and storage module in completing their respective initialization work; the initialization of the control module itself is also carried out simultaneously, including task thread resource pooling, task management initialization, and log function initialization. Task scheduling: First, the task is decomposed. The control module generates specific filtering conditions based on the configuration and divides the overall data into corresponding task subsets. A globally unique ID is generated for each task and a task execution object is created. The storage module is called to generate a separate storage space for the task. Then, a multi-threaded task execution process is initiated, and the thread pool and task module are scheduled to perform analysis jobs. During task execution, this module is responsible for monitoring the running status of the initiated task and maintaining its running information; Task Management: A task list is maintained in memory, recording the initial, running, and result information of each initiated task. The lifecycle information of each task, including task ID, condition settings, execution results, and exception information, is synchronized to the task list. After a single task is completed, this function synchronizes the corresponding information in the task list to the storage area. When all tasks are completed, this function organizes and aggregates all entries in the list to generate a unified execution report.

5. A method for periodic analysis of multi-condition time series data based on AI technology, characterized in that, The system for multi-condition time-series data periodic analysis based on AI technology, as described in any one of claims 1-4, includes the following steps: Create a global configuration file containing input / output configuration, condition configuration, preprocessing configuration, and analysis configuration; During system initialization, the control module calls the task module, AI analysis module, and storage module to complete their respective initialization tasks; the control module itself is also initialized simultaneously, including task thread resource pooling, task management initialization, and log function initialization. The control module reads the condition configuration information in the configuration, splits it into multiple tasks according to the configuration, and divides and filters the dataset according to the filtering conditions; it synchronously creates a task list, generates a unique ID for each combination of conditions, and creates a task execution object and an independent directory for each task; for each task, it allocates thread resources to the task in batches and executes them in the order of data processing, AI analysis, and task information integration. The task management function of the control module receives the result object returned by each task, updates the task list and writes it to the dedicated task storage area directory; when all tasks are completed, a unified execution report is generated and stored in the task storage area.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the multi-condition time series data periodic analysis method based on AI technology as described in claim 5.

7. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the multi-condition time-series data periodic analysis method based on AI technology as described in claim 5.

Images