Automatic test method and system suitable for large-current program control

Abstract

The invention relates to the technical field of electrical performance testing, in particular to an automatic testing method and system suitable for large-current program control. The method comprises the following steps: obtaining a large-current programmed control system log and large-current programmed control system monitoring data, and carrying out system state modeling to obtain a system state dynamic model; performing workflow environment up-sampling on the system state dynamic model to obtain scene simulation data; performing simulation state mode recognition on the system state dynamic model according to the scene simulation data so as to obtain system process simulation state mode data; performing system state mode score calculation on the system process simulation state mode data based on a system state constraint dynamic model to obtain programmed control system state score data; according to the program control system state score data and the large-current program control system log, process control system repair strategy selection is carried out, and therefore a system repair strategy is obtained. The stability of the large-current programmed control system can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of electrical performance testing, and in particular to an automated testing method and system suitable for large current process control. Background Art

[0002] Large-scale process control systems manage and monitor complex processes in an automated way, ensuring the stability and efficiency of the production process. However, in the design and application of these systems, how to ensure their reliability, stability and safety, especially their accuracy in complex working environments, has always been the focus of the industry. In order to ensure the performance and stability of these control systems, automated testing has become a crucial technical means. Traditional testing methods for large-scale process control systems can be roughly divided into manual testing, script-based testing, and simulation testing. Although these traditional testing methods can work in some cases, they have many defects that cannot be ignored. The manual testing method is one of the earliest methods used in traditional testing. Testers manually operate the control system, observe its response and record test data. However, manual testing usually relies on manual judgment, which is prone to human errors during the test process, and the test efficiency is low and the reproducibility is poor. Especially when facing large-scale systems or complex processes, manual testing is difficult to cover all test scenarios, and it is difficult to ensure the stability and repeatability of test results. Summary of the invention

[0003] Based on this, it is necessary for the present invention to provide an automated testing method and system suitable for large current process control to solve at least one of the above technical problems.

[0004] To achieve the above purpose, an automated testing method suitable for large current process control includes the following steps: Step S1: Obtain the log of the large electric process control system and the monitoring data of the large electric process control system, and perform workflow topology modeling according to the log of the large electric process control system, so as to obtain the workflow topology model; perform system state modeling according to the workflow topology model and the monitoring data of the large electric process control system, so as to obtain the system state dynamic model; Step S2: mapping the system state dynamic model to the requirements of the program-controlled system, thereby obtaining the system state constraint dynamic model, and performing real-time system state anomaly detection based on the system state constraint dynamic model, thereby obtaining system abnormal state data; performing workflow environment upsampling based on the system abnormal state data and the system state dynamic model, thereby obtaining scenario simulation data; Step S3: performing program-controlled system state simulation on the system state dynamic model according to the scenario simulation data, thereby obtaining program-controlled system simulation state data, and performing simulation state pattern recognition based on the program-controlled system simulation state data, thereby obtaining system process simulation state pattern data; Step S4: Calculate the system state mode score of the system process simulation state mode data based on the system state constraint dynamic model, thereby obtaining the program control system state score data; Step S5: Select a process control system repair strategy based on the program control system status score data and the large-scale program control system log, thereby obtaining a system repair strategy, and upload it to the large-scale program control system to execute the program repair task.

[0005] The present invention can greatly improve the reliability, stability and safety of large-scale electric process control systems in complex working environments by constructing accurate models and simulation mechanisms. By acquiring and analyzing the log data and monitoring data of the large-scale electric process control system, the modeling of the workflow topology structure can be realized first, and then the structural relationship and dynamic characteristics of the large-scale electric process control system can be clarified, ensuring that the state of the large-scale electric process control system can be accurately monitored and understood during the test and operation process. Based on this model, the dynamic state characteristics of the large-scale electric process control system can be further obtained to ensure that the large-scale electric process control system can always maintain a predetermined stable state in a changing environment. Through the dynamic modeling and demand mapping of the state of the large-scale electric process control system, the abnormal state can be captured in real time during the operation of the large-scale electric process control system, so as to timely discover potential large-scale electric process control system failures. This real-time abnormal detection and data acquisition provide high-quality data support for subsequent large-scale electric process control system simulation and repair, ensuring the comprehensiveness and accuracy of the test. With the generation of scene simulation data, the operating state of the large-scale electric process control system in different environments can be simulated, which lays the foundation for further state simulation and pattern recognition, and can obtain more abundant system behavior data through simulation. The identification of simulation state patterns not only helps to understand the different possible states of the system operation, but also reveals the relationship between potential process patterns and states, providing a strong basis for subsequent performance optimization and fault prevention. The state pattern score calculation based on the state constraint model of the large-scale electric process control system can quantitatively evaluate the overall state of the large-scale electric process control system, and further provide data support for the maintenance and optimization of the large-scale electric process control system. Through these scores, the performance of the large-scale electric process control system can be accurately evaluated, abnormal performance can be discovered in time, and a scientific basis can be provided for the formulation of effective repair strategies. This data-driven repair strategy selection not only improves the stability of the process control system, but also reduces manual intervention through automation, reduces the possibility of human errors, and thus significantly improves the reliability and safety of the entire system. Overall, these steps are combined to achieve accurate monitoring and efficient testing of the system in a complex environment, can automatically detect and repair system anomalies, and ensure the long-term and efficient operation of the large-scale electric process control system while ensuring the stability of the production process. The core advantage of this method is that it provides an automated, efficient, and repeatable test and repair solution through data-driven modeling and simulation, reduces the need for manual intervention, and improves the reliability and maintainability of the overall large-scale electric process control system.

[0006] Optionally, step S1 specifically includes: Step S11: Obtaining the large current process control system log and the large current process control system monitoring data; Step S12: performing data preprocessing on the large electric process control system log and the large electric process control system monitoring data respectively, so as to obtain the process control system log to be analyzed and the process control system monitoring data to be analyzed; Step S13: integrating system function module features according to the program-controlled system log to be analyzed, thereby obtaining system function module data; Step S14: Modeling the workflow topology structure based on the system function module data, thereby obtaining a workflow topology structure model; Step S15: System state modeling is performed according to the workflow topology model and the large power process control system monitoring data, so as to obtain a system state dynamic model.

[0007] The present invention can effectively improve the transparency and monitoring efficiency of system operation by gradually analyzing and modeling the logs and monitoring data of the large-scale electric process control system. By obtaining the logs and monitoring data of the system, real and comprehensive raw data are provided for subsequent analysis. This process lays a solid foundation for the data preprocessing stage and ensures that the data required for analysis are accurate, complete and representative. During the data preprocessing process, the system can remove noise data, fill in missing values, and perform standardization processing, thereby ensuring the quality of the data to be analyzed and providing reliable data support for subsequent steps. The system function module feature integration can extract the key function module information of the system from the complex log data, laying the foundation for building a more accurate model. Through this feature integration, the system can more clearly identify the correlation and dependency between different function modules, thereby providing strong support for the structural analysis of the entire system. In the further workflow topology modeling process, through the analysis of the system function modules, the collaborative relationship and information flow of each part of the system can be accurately constructed. This model not only reveals the internal organizational structure of the system, but also can intuitively present the workflow and interaction mode between different modules, providing an important basis for system optimization and troubleshooting. Dynamic modeling of system status based on topological structure model and monitoring data can realize real-time monitoring and dynamic prediction of system status. This dynamic model can reflect the behavior changes of the system under different operating conditions, help managers to identify potential problems in time and issue early warnings, and effectively improve the safety and stability of the system.

[0008] Optionally, step S14 is specifically: Step S141: extracting function module interaction events and function module communication from system function module data, thereby obtaining function module interaction data and function module communication data; Step S142: Connecting the function module communication nodes of the program control system according to the function module communication data, thereby obtaining the function module communication node connection data; Step S143: performing a function module interaction event time series analysis on the function module interaction data, thereby obtaining the function module interaction event time series data; Step S144: identifying time series dependency relationships based on the time series data of the functional module interaction events, thereby obtaining functional module dependency relationship data; Step S145: Modeling the workflow topology structure according to the functional module communication node connection data and the functional module dependency data, thereby obtaining a workflow topology structure model.

[0009] The present invention can effectively identify the interaction and communication mode between each functional module by extracting and analyzing the system function module data, thereby providing data support for subsequent system optimization and troubleshooting. First, extracting the functional module interaction events and communication data can help clarify the actual data flow and interaction between modules, and then reveal potential efficiency bottlenecks or communication problems. Through further analysis of the communication data, the connection mode and dependency relationship between each functional module can be identified, providing a clear framework for the optimal design of the program-controlled system. At the same time, time series analysis and dependency identification are carried out to help reveal the temporal association and work dependency between each module, and further enhance the stability and reliability of the system under different load conditions. Based on these analysis results, a workflow topology model is established, which not only helps to fully understand the overall workflow of the system, but also can provide strong support for the optimization, automatic scheduling and fault prediction of the system, thereby improving the operating efficiency of the system and the ability to cope with complex tasks.

[0010] Optionally, step S15 is specifically: Step S151: define the state space based on the workflow topology model, so as to obtain the system state space; Step S152: extracting system operation data from the large-scale process control system log, thereby obtaining system operation data, and defining state transition rules for the system operation data and the large-scale process control system monitoring data, thereby obtaining state transition rules; Step S153: performing state transfer matrix statistics according to the state transfer rule and the system state space, thereby obtaining the system state transfer matrix; Step S154: Perform time series dynamic modeling based on the system state transfer matrix and the large-scale power process control system monitoring data to obtain a system state dynamic model.

[0011] The present invention can accurately identify possible operating states and state change ranges by constructing a workflow topology model and defining the state space based on the model, providing a comprehensive framework for subsequent state analysis. This process helps to quantify the entire process, thereby clarifying the working mode and state transition under different conditions, which is helpful for further dynamic monitoring and optimization. On this basis, the system operation data is extracted and the state transition rules are defined for it and the monitoring data, which can refine the logical relationship between each operation and state change and reveal the dynamic evolution mechanism of each link in the system. The definition of this rule not only provides a basis for accurate modeling, but also provides basic data support for later state prediction and fault diagnosis. By combining the state transition rules with the state space, a state transition matrix is ​​formed. The matrix can accurately describe the conversion relationship between different states, provide the probability and conditions of the transition between system states, and provide a systematic perspective for dynamic monitoring. Based on this matrix, combined with the monitoring data, time series dynamic modeling can capture the trend and law of state changes in real time, helping managers to respond promptly when the system state changes. Time series dynamic modeling not only helps to understand the past behavior of the system, but also predicts the possible future state of the system, providing a powerful tool for the long-term stability, early warning mechanism and troubleshooting of the system. Through this series of steps, the monitoring capabilities of the entire system have been greatly improved, which can more efficiently identify potential problems, optimize resource allocation, and improve workflow automation and system reliability.

[0012] Optionally, step S2 specifically includes: Step S21: acquiring performance index data of a large current process control system, and performing system demand statistics on the performance index data of the large current process control system, thereby obtaining demand data of the large current process control system; Step S22: mapping the program control system requirements to the system state dynamic model according to the large electric program control system requirement data, thereby obtaining the system state constraint dynamic model; Step S23: acquiring the real-time monitoring data of the large-scale electric process control system, and performing real-time system state abnormality detection on the real-time monitoring data of the large-scale electric process control system according to the system state constraint dynamic model, thereby obtaining system abnormal state data; Step S24: integrating the system operation process structure of the system state dynamic model, thereby obtaining a system operation process structure model; Step S25: Perform workflow environment upsampling according to the system abnormal state data and the system operation process structure model, so as to obtain scenario simulation data.

[0013] The present invention can effectively understand the operation requirements of the system by acquiring and counting the performance index data of the large-scale electric process control system, and provide detailed system requirement information for subsequent analysis. These requirement data can not only reflect the performance standards of the system in normal operation, but also provide data support for system optimization. Based on these requirement data, they are mapped to the system state dynamic model, and a system state constraint dynamic model can be constructed. This model provides more clear constraints for the operation of the system, ensuring that the system operation is carried out within a predetermined demand range, thereby improving the stability and reliability of the system. Through the acquisition and abnormality detection of real-time monitoring data, potential problems and state anomalies in the system can be identified in real time to ensure that the system can respond to and handle emergencies in a timely manner. This process can not only detect and locate faults in advance, but also provide important data basis for troubleshooting and system recovery, thereby reducing system downtime and production losses. By integrating the system state dynamic model and the operation process structure, the overall architecture of the system operation can be further sorted out to help analyze the association and dependence between each link. This integration process provides a clear framework for more efficient system scheduling, operation optimization and process reconstruction. Using abnormal state data and the system operation process structure model to perform environmental upsampling and generate scene simulation data is helpful for multi-scenario simulation and testing of the system under different operating conditions. Scenario simulation can help analyze the performance and response of the system under various complex environments and operating conditions, thereby providing reliable decision support, identifying potential risks in advance, and optimizing system design and workflow, further improving the system's emergency response capabilities and overall operational efficiency. Through these steps, the monitoring, optimization and fault management capabilities of the entire large power process control system have been significantly improved, providing strong support for achieving efficient and stable power process control.

[0014] Optionally, step S23 is specifically: Step S231: acquiring real-time monitoring data of a large electric process control system, and performing data preprocessing on the real-time monitoring data of the large electric process control system, thereby obtaining real-time monitoring data of the system to be analyzed; Step S232: predicting the real-time dynamic state of the system to be analyzed based on the system state constraint dynamic model, thereby obtaining the real-time dynamic state data of the system; integrating the performance indicator data of the large electric process control system based on the system state constraint dynamic model, thereby obtaining the expected dynamic state data of the system; Step S233: calculating the system expected dynamic state deviation amount for the system real-time dynamic state data and the system expected dynamic state data, thereby obtaining the system dynamic state deviation amount data; Step S234: performing nonlinear deviation state identification on the system real-time dynamic state data according to the system dynamic state deviation data, thereby obtaining system abnormal state data.

[0015] The present invention makes the data clearer and more standardized through the preprocessing of real-time monitoring data, and provides accurate basic information for subsequent analysis. Then, by predicting the real-time dynamic state through a dynamic model based on system state constraints, the current operating state of the system can be accurately captured, and the understanding and control of the actual situation of the system can be improved. At the same time, the expected dynamic state integration of performance indicator data is helpful to comprehensively evaluate the working performance of the system, and then provide a basis for system optimization and adjustment. By calculating the deviation between the real-time dynamic state of the system and the expected dynamic state, the deviation of the system can be quantified, so as to more accurately monitor the abnormalities that may occur during the operation of the system. Through nonlinear deviation state identification, various abnormal states that occur in the actual operation of the system can be effectively identified, so as to timely warn and take corresponding measures to avoid serious failures of the system. The implementation of this series of steps can greatly improve the stability and reliability of the system, reduce system downtime, optimize resource allocation, and thus improve overall production efficiency and economic benefits.

[0016] Optionally, step S25 is specifically: Step S251: generating an abnormal operating system working scenario for the system operation process structure model based on the system abnormal state data, thereby obtaining an abnormal workflow scenario data set; generating a normal operating system working scenario for the system operation process structure model based on the system expected dynamic state data, thereby obtaining a normal workflow scenario data set; Step S252: performing Monte Carlo upsampling processing on the abnormal workflow scenario data set and the normal workflow scenario data set, thereby obtaining a random workflow scenario data set; Step S253: parameterizing the work scenario based on the system operation process structure model, thereby obtaining a work scenario parameter set; Step S254: Perform workflow scenario simulation on the random workflow scenario data set and the workflow scenario parameter set to obtain scenario simulation data.

[0017] The present invention can comprehensively reflect the performance of the system under different working conditions by generating data sets of system working scenarios of abnormal operation and normal operation. This helps to analyze the operation of the system from different angles, especially the work process under abnormal conditions, and provides valuable data support for the optimization and fault diagnosis of the system. Then, through Monte Carlo upsampling processing, the diversity of the data set can be enhanced by randomization methods, making the data set richer, thereby improving the robustness and adaptability of model training, avoiding overfitting, and being able to better simulate complex and random operation scenarios in reality. In addition, the parameterization of the working scenario can simplify and structure the complex workflow, provide a clear parameter framework for subsequent simulation and analysis, and ensure the efficiency and accuracy of the simulation process. Workflow scenario simulation can convert theoretical data into simulation data in practical applications, provide sufficient experimental support for system verification, testing and optimization, make the performance evaluation of the system more accurate, and be able to discover potential problems in time and improve them.

[0018] Optionally, step S3 specifically includes: Step S31: extracting historical working scene features according to the program control system monitoring data to be analyzed, thereby obtaining historical working scene data; Step S32: based on the system state dynamic model, the system historical state backtracking is performed on the program-controlled system log to be analyzed, so as to obtain historical system state data; Step S33: performing correlation analysis on the historical working scene data and the historical system status data, thereby obtaining scene-state correlation data; Step S34: simulating the state of the program-controlled system on the system state dynamic model according to the scene-state correlation data and using the scene simulation data, thereby obtaining the program-controlled system simulation state data; Step S35: Perform simulation state pattern recognition based on the program control system simulation state data, thereby obtaining system process simulation state pattern data.

[0019] The present invention can help to deeply understand the working state and operation characteristics of the program-controlled system in the past period of time by extracting the characteristics of historical working scenes, and provide detailed background data for subsequent analysis. By combining the system state dynamic model to trace back the historical state of the program-controlled system log, the historical operation state of the system can be accurately restored to help identify potential operation problems and performance bottlenecks. The collection and traceback of this historical data provides a comprehensive perspective for subsequent analysis. The correlation analysis of historical working scene data and historical system state data can reveal the intrinsic connection between the system and the system state in different working scenes, help identify which working conditions and states are most likely to cause abnormalities or performance degradation, and thus provide guidance for optimizing the system. These scene-state correlation data provide a scientific basis for subsequent simulation and optimization models. Using these correlation data and scene simulation data to simulate the system state dynamic model can generate more accurate system state simulation data, thereby improving the prediction ability of the model, so that the system can operate more efficiently and reliably when facing future working scenes. Based on the simulation state data, pattern recognition can be performed to extract the simulation state mode of the system process, thereby identifying the potential risks and optimization space of the system, which helps to warn the abnormal state of the system in advance and avoid possible failures and performance degradation. The comprehensive application of this series of steps can not only improve the reliability, stability and efficiency of system operation, but also provide more accurate decision-making support for the long-term maintenance and optimization of the system.

[0020] Optionally, step S35 is specifically: Step S351: setting the system state mode category based on the large current process control system performance indicator data, thereby obtaining the system state mode category data; Step S352: extracting simulation state features from the program control system simulation state data, thereby obtaining simulation state feature data; Step S353: performing simulation state similarity calculation according to the simulation state feature data, thereby obtaining simulation state feature similarity data; Step S354: performing simulation state clustering calculation on the program control system simulation state data based on the simulation state feature similarity data, thereby obtaining the program control system clustered simulation state data; Step S355: Perform simulation state pattern recognition on the clustered simulation state data of the program-controlled system according to the system state pattern category data, thereby obtaining system process simulation state pattern data.

[0021] Optionally, the present specification further provides an automated testing system applicable to large current program control, which is used to execute the automated testing method applicable to large current program control as described above, and the automated testing system applicable to large current program control includes: The system state modeling module is used to obtain the log of the large-scale electric process control system and the monitoring data of the large-scale electric process control system, and to model the workflow topology structure according to the log of the large-scale electric process control system, so as to obtain the workflow topology structure model; and to model the system state according to the workflow topology structure model and the monitoring data of the large-scale electric process control system, so as to obtain the system state dynamic model; The system state anomaly detection module is used to map the system state dynamic model to the requirements of the program-controlled system, thereby obtaining the system state constraint dynamic model, and to perform real-time system state anomaly detection based on the system state constraint dynamic model, thereby obtaining system abnormal state data; and to perform workflow environment upsampling based on the system abnormal state data and the system state dynamic model, thereby obtaining scenario simulation data; The program-controlled system state simulation module is used to simulate the program-controlled system state on the system state dynamic model according to the scenario simulation data, thereby obtaining the program-controlled system simulation state data, and to perform simulation state pattern recognition based on the program-controlled system simulation state data, thereby obtaining the system process simulation state pattern data; A mode score calculation module is used to calculate the system state mode score of the system process simulation state mode data based on the system state constraint dynamic model, so as to obtain the program control system state score data; The repair strategy selection module is used to select the process control system repair strategy based on the program control system status score data and the large-scale program control system log, so as to obtain the system repair strategy and upload it to the large-scale program control system to execute the program repair task.

[0022] The automated testing system for large current program control of the present invention can implement any automated testing method for large current program control of the present invention, and is used to combine the operation and signal transmission medium between various modules to complete the automated testing method for large current program control. The internal modules of the system cooperate with each other, thereby improving the stability of the large current program control system and the test efficiency of the program control test. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments thereof made with reference to the following drawings: Figure 1 It is a schematic diagram of the steps of the automated testing method applicable to large current process control of the present invention; Figure 2 Detailed step flow diagram of step S1 in the present invention; Figure 3 Detailed step flow diagram of step S2 in the present invention; The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0024] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0025] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0026] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0027] To achieve this, please refer to Figures 1 to 3 The present invention provides an automated testing method suitable for large current process control, the method comprising the following steps: Step S1: Obtain the log of the large electric process control system and the monitoring data of the large electric process control system, and perform workflow topology modeling according to the log of the large electric process control system, so as to obtain the workflow topology model; perform system state modeling according to the workflow topology model and the monitoring data of the large electric process control system, so as to obtain the system state dynamic model; In this embodiment, by integrating with the large power process control system (such as SCADA system) in the power system, real-time control data, equipment operation logs and alarm information are automatically extracted. These data include the operating status of the equipment, sensor data, control instructions, fault records, etc. Use API or data interface to obtain this information regularly (for example, every hour, every day) and store it in the central database for subsequent analysis. According to the log data of the large power process control system, a workflow topology model of the power system is constructed. The topological structure includes information such as the connection relationship, control logic and fault transfer path between various devices. The topological structure of the process is represented by a graph theory method (such as a directed graph), in which the node represents the device or control node, and the edge represents the signal transmission or control relationship between the devices. Assuming that in a power grid control system, a topological model can be used to represent how generators, transformers, distribution equipment, etc. are connected and interacted with each other to ensure that each device and system can work correctly. Using the information obtained from the topological model and real-time monitoring data (such as equipment temperature, current, voltage, load, etc.), a dynamic model of the system state is constructed. The model can reflect the operating status of the system, reflect the health status of the power system at any time, dynamically adjust the operating status of the equipment, and simulate different operating scenarios. For example, a system status dynamic model may include variables such as the system's real-time load forecast, equipment health, and energy flow paths.

[0028] Step S2: mapping the system state dynamic model to the requirements of the program-controlled system, thereby obtaining the system state constraint dynamic model, and performing real-time system state anomaly detection based on the system state constraint dynamic model, thereby obtaining system abnormal state data; performing workflow environment upsampling based on the system abnormal state data and the system state dynamic model, thereby obtaining scenario simulation data; In this embodiment, the requirements of the large power process control system are mapped to the system state dynamic model. That is, by comparing the control system's goals and the current system state, it is identified which states do not meet the expected requirements. For example, if the system requires that the load of a certain transformer does not exceed a certain threshold, but the system state shows that its load has reached or exceeded the threshold, then this state can be marked as abnormal. When the system is running, the state of each device is monitored in real time by comparing with the predetermined threshold, and abnormality detection is performed. Data-driven abnormality detection methods, such as statistical control charts or machine learning-based abnormality detection models, are used to timely identify abnormal states in the system. For example, when the current of the transformer exceeds the set safety threshold, the system automatically identifies it as an abnormal state and generates an alarm. Based on the identified abnormal state data, possible working scenario data is generated. By upsampling the data of these abnormal states, the possibility of abnormal scenarios is expanded. For example, when the current of a certain device is abnormal, different load changes and system responses can be simulated to generate a series of new simulation data.

[0029] Step S3: performing program-controlled system state simulation on the system state dynamic model according to the scenario simulation data, thereby obtaining program-controlled system simulation state data, and performing simulation state pattern recognition based on the program-controlled system simulation state data, thereby obtaining system process simulation state pattern data; In this embodiment, based on the obtained scenario simulation data, the state of the power system is simulated through simulation software (such as MATLAB, Simulink, etc.), simulating the system response under different faults, equipment aging or load fluctuations. For example, the simulation system will simulate how the system automatically switches to backup equipment to ensure stable operation of the power grid after a transformer failure. Pattern recognition is performed on the simulation data, and machine learning algorithms (such as cluster analysis, support vector machines, neural networks, etc.) are used to identify different state modes from the simulation data. For example, the simulation data will show different working states such as "normal operation mode", "overload mode", and "equipment failure mode" to help identify potential problems.

[0030] Step S4: Calculate the system state mode score of the system process simulation state mode data based on the system state constraint dynamic model, thereby obtaining the program control system state score data; In this embodiment, the state mode score is calculated based on the simulation state mode data and the system state constraint dynamic model. The score of each mode is calculated by evaluating the impact of different state modes on the overall stability, reliability and safety of the system. Using, for example, weighted method, fuzzy logic or optimization algorithm, the degree of impact of each state mode is converted into a score to evaluate its impact on the system. For example: If a certain mode (such as overload mode) causes the system to fail frequently in simulation, then the score of this mode will be low, indicating that it poses a greater threat to system stability. By combining the score data of multiple state modes, the current overall state score of the system is calculated. This score reflects the health status, operating efficiency and risk level of the system.

[0031] Step S5: Select a process control system repair strategy based on the program control system status score data and the large-scale program control system log, thereby obtaining a system repair strategy, and upload it to the large-scale program control system to execute the program repair task.

[0032] In this embodiment, a suitable repair strategy is selected based on the status score of the program control system and the abnormal data in the system log. The repair strategy includes adjusting control parameters, switching backup equipment, increasing load balancing, enabling redundant lines, etc. For example, when it is detected that the load of a certain transformer is too high, choose to start the backup transformer to share the load, or adjust the distribution network to reduce the load in the area. After the repair strategy is selected, the repair plan will be automatically uploaded to the large power program control system, and the repair task will be executed according to the uploaded control instructions. For example, the large power program control system will automatically issue control instructions to adjust the workload of the transformer or switch to the backup system. During the execution process, the system monitors the repair effect in real time, and rolls back or tries again when the repair fails.

[0033] Optionally, step S1 specifically includes: Step S11: Obtaining the large current process control system log and the large current process control system monitoring data; In this embodiment, the log information and monitoring data of the large power process control system are obtained through the system management platform or data acquisition terminal. Specifically, the Industrial Internet of Things (IIoT) platform can be used to connect to various devices and sensors in the large power process control system to collect the operation log and real-time monitoring data of the system. Log information usually includes equipment running time, alarm records, fault information, etc.; monitoring data includes the operating status of power equipment, data collected by various sensors (such as current, voltage, temperature, pressure, etc.). Standard data interfaces (such as Modbus, OPC-UA, etc.) can be used to pull data to ensure the timeliness and integrity of the data. The communication protocol between the data acquisition system (such as the SCADA system) and each device can be configured to obtain the device status and data log every 5 minutes. Use the REST API interface to extract real-time data (such as voltage, current, load, etc.) from different monitoring modules and save them as structured log files or database records.

[0034] Step S12: performing data preprocessing on the large electric process control system log and the large electric process control system monitoring data respectively, so as to obtain the process control system log to be analyzed and the process control system monitoring data to be analyzed; In this embodiment, data preprocessing is to ensure that subsequent analysis can be performed on clean and structured data. For system log data, the original log can be cleaned by regular expressions or log parsing tools (such as Logstash) to remove irrelevant noise information and extract key information (such as event type, timestamp, device ID, etc.). For monitoring data, data interpolation algorithms (such as linear interpolation, K nearest neighbor interpolation, etc.) can be used to fill in part of the data lost due to communication failure or sensor failure, and detect and correct outliers (such as outlier detection based on Z-score). Use Python to write a data cleaning program, preprocess logs and monitoring data through the pandas library, remove useless fields and interpolate missing values. Classify system logs, mark events by severity, and store them in a format that is easy to analyze (such as CSV or JSON). Use a smoothing algorithm (such as Savgol smoothing method) to smooth the monitoring data to reduce the impact of fluctuations caused by noise.

[0035] Step S13: integrating system function module features according to the program-controlled system log to be analyzed, thereby obtaining system function module data; In this embodiment, the system log is a record describing the operation history of the large power program control system, including various events that occurred in the system. The program control system log to be analyzed is analyzed, and the log data is grouped according to functional modules (such as power monitoring, equipment control, alarm management, etc.). Through event timing analysis, the activity characteristics of each functional module in a specific time period can be extracted (such as the execution of operation commands, the response to alarms, etc.). Feature extraction is performed on the log information of each functional module, for example, common event types are extracted through frequency analysis, and the association relationship between functional modules is found through association rule analysis. Use machine learning models (such as decision trees or random forests) to classify log data and extract the characteristics of different functional modules such as device control modules, monitoring modules, and alarm modules. The features are integrated into performance indicator data for each functional module, such as failure rate, response time, number of equipment startups, etc., and saved in the database for subsequent analysis.

[0036] Step S14: Modeling the workflow topology structure based on the system function module data, thereby obtaining a workflow topology structure model; In this embodiment, the workflow topology model of the system is constructed according to the data of each functional module and the interaction between them. Specifically, a graph theory method can be used to regard each functional module in the system as a node of the graph, and the dependency relationship between the functional modules as the edge of the graph. Use deep analysis technology (such as co-occurrence analysis) to find out the interaction between the modules, and then build a topological structure diagram reflecting the information flow, control flow, and event flow between the modules. The state transition law of each module can be further modeled using a Bayesian network or a Markov decision process (MDP). A graphical visualization tool such as Gephi can be used to draw a workflow topology diagram between functional modules, where nodes represent functional modules and edges represent the relationship between them. Based on the topological diagram, the bottleneck of the workflow is analyzed and an optimization scheme is proposed, such as adding redundant modules, optimizing data transmission paths, etc.

[0037] Step S15: System state modeling is performed according to the workflow topology model and the large power process control system monitoring data, so as to obtain a system state dynamic model.

[0038] In this embodiment, the dynamic state model of the system is established by combining the workflow topology model and monitoring data. The state change path and dependency relationship of each module in the system are determined through the topological relationship of the workflow. Based on real-time monitoring data (such as device status, sensor data, etc.), the state of the system is modeled using a state space model (such as a Markov chain) or a time series data modeling method (such as a long short-term memory network LSTM). Dynamic models can be used to simulate changes in system states under different circumstances and predict possible system failures or performance degradation. The LSTM model can be used to model monitoring data, predict state changes of each module in the future time period, and determine whether there is a risk of failure. According to the workflow topology and monitoring data, weights are assigned to each node in the graph model to simulate the behavior of the system under different loads and operating modes, thereby achieving dynamic prediction of the health status of the system.

[0039] Optionally, step S14 is specifically: Step S141: extracting function module interaction events and function module communication from system function module data, thereby obtaining function module interaction data and function module communication data; In this embodiment, interaction events and communication data are extracted from different functional modules of the system. For example, in an automated production control system, the functional modules may include sensor modules, actuator modules, monitoring modules, etc. These modules exchange data through specific communication protocols (such as Modbus, CAN, Ethernet, etc.). By analyzing the system log file to capture the system communication, the interaction event data between different functional modules can be extracted. Interaction events include actions such as signal transmission and data exchange between modules. The extracted communication data may include the timestamp of each event, the event type (such as read, write, request response, etc.), and the functional modules involved.

[0040] Step S142: Connecting the function module communication nodes of the program control system according to the function module communication data, thereby obtaining the function module communication node connection data; In this embodiment, based on the extracted functional module communication data, a communication node connection diagram of each functional module in the system is constructed. Taking a production system controlled by a PLC (programmable logic controller) as an example, the functional modules of the control system include sensors, actuators, and central control units. By analyzing the communication data, the communication nodes of each functional module and the connection relationship between them can be determined. For example, assuming that the sensor module communicates with the central control unit via Ethernet, and the actuator module is connected to the central control unit via a serial port, then in the communication node connection diagram, the sensor, central control unit, and actuator are respectively used as nodes, and the sensor to the central control unit and the central control unit to the actuator are communication connections. The data finally generated will include the identifier of each communication node and the connection mode between the nodes (such as point-to-point, broadcast, cluster, etc.).

[0041] Step S143: performing a function module interaction event time series analysis on the function module interaction data, thereby obtaining the function module interaction event time series data; In this embodiment, a time series analysis is performed on the interaction event data extracted in the previous step. Assume that in an industrial automation control system, there is a certain time delay between sensor data acquisition and actuator action. The time relationship and order of event occurrence can be analyzed by sorting and counting the interaction data. For example, by analyzing the timestamp, the time interval between each interaction event (such as the acquisition time of sensor data and the control command time of the actuator) is determined, thereby obtaining a clear time series data set. These time series data will reflect the order and timeliness of the interaction between functional modules and reveal the working dependencies between modules.

[0042] Step S144: identifying time series dependency relationships based on the time series data of the functional module interaction events, thereby obtaining functional module dependency relationship data; In this embodiment, the dependency between modules is further identified based on the time series data of the interaction events of the functional modules. For example, by analyzing the event sequence of each module, it can be found that the output of some modules depends on the input of other modules. In a robot control system, the input data of the sensor (such as temperature sensor, pressure sensor) depends on the state of the actuator (such as motor speed or position). Through the correlation analysis of the time series data, the dependency between the sensor module and the actuator module can be identified. At this time, the methods such as association rules and time window analysis can be used to obtain which modules have events in sequence and the sequence dependency between them, thereby obtaining the dependency data of the functional modules.

[0043] Step S145: Modeling the workflow topology structure based on the functional module communication node connection data and the functional module dependency data, thereby obtaining a workflow topology structure model.

[0044] In this embodiment, the functional module communication node connection data and dependency data obtained above are used to build the workflow topology model of the system. Taking a complex automated production line as an example, there are multiple functional modules in the system, each module exchanges data through a specific communication node, and there is a temporal dependency. Based on these data, a graph theory method can be used to take the various functional modules of the system as nodes of the graph, the communication connections between modules as edges of the graph, and the dependencies as constraints of the graph. In this way, a clear topology model can be constructed to show the communication paths, data flows, and dependency sequences of each functional module. This topology model can provide an important basis for system optimization, fault diagnosis, and performance improvement.

[0045] Optionally, step S15 is specifically: Step S151: define the state space based on the workflow topology model, so as to obtain the system state space; In this embodiment, based on the structure of the topological model, the state of each node is discretized and defined. For example, for a substation, "normal operation", "fault" or "under maintenance" can be defined as the state. In this process, it is necessary to ensure that the defined state space covers all possible operation or fault conditions in the system. The output of this step is the state space of the system, which is a multidimensional space that contains the states of all possible equipment or operation combinations. Assuming that a power generation system includes a generator, a substation, and a distribution equipment, the state space can be defined as a set of discrete states of each device. For example, the state space of the generator is {normal operation, outage, fault}, the state space of the substation is {operation, outage, pending repair}, and the state space of the distribution equipment is {normal, overload, short circuit}. By combining these states, the state space of the entire system can be obtained.

[0046] Step S152: extracting system operation data from the large-scale process control system log, thereby obtaining system operation data, and defining state transition rules for the system operation data and the large-scale process control system monitoring data, thereby obtaining state transition rules; In this embodiment, operation logs are extracted from the large power process control system, and these logs record the real-time state changes of each device or operation in the system. Specifically, the log contains information such as timestamp, device ID, operation type (such as start and stop, load adjustment), operation result (such as success, failure), etc. These operation data are extracted through log parsing algorithms (such as regular expressions or time series analysis methods). By analyzing the extracted system operation data and monitoring data, state transition rules can be defined. These rules describe the conditions from one state to another. For example, if the operation of a certain device (such as startup) is successful, the system state changes from "shutdown" to "running" state. If the operation fails, the state rolls back or enters the "fault" state. For example, in the log of the generator set, the following transitions can be observed: from the state "generator shutdown" to "generator startup", the condition is that the device operation signal is startup and there is no fault report. From the state "generator startup" to "generator operation", the condition is that the device reaches certain stable operating parameters after startup. In the extracted log, it is assumed that the operation log of a substation shows that a maintenance operation is successful, resulting in a change in state from "pending repair" to "normal operation". This event can be used as part of the state transition rule. If no fault is found during the next maintenance of the same substation, the state remains as "normal operation"; if a fault occurs, the state changes to "fault" state.

[0047] Step S153: performing state transfer matrix statistics according to the state transfer rule and the system state space, thereby obtaining the system state transfer matrix; In this embodiment, based on the state space and state transition rules defined above, a state transition matrix can be constructed. The state transition matrix is ​​a matrix that represents the probability of state change, in which each element represents the probability of transition from one state to another. According to the log data, the frequency of transition from one state to another is counted. For example, by analyzing the operation log of the system, the frequency of the state from "normal operation of the generator" to "generator shutdown" can be recorded, and these frequencies can be converted into transition probabilities. By statistics of multiple operations and state changes, a complete state transition matrix is ​​gradually constructed. Assume that by analyzing the historical data of the large power process control system, it is found that when the system state transfers from "normal operation" to "fault", the transfer frequency is 30%; the frequency of transfer from "normal operation" to "maintenance" state is 20%; the remaining 50% of the transfers are maintained in the "normal operation" state. According to these statistics, the corresponding state transfer matrix is ​​formed.

[0048] Step S154: Perform time series dynamic modeling based on the system state transfer matrix and the large-scale power process control system monitoring data to obtain a system state dynamic model.

[0049] In this embodiment, based on the constructed state transition matrix and real-time monitoring data (such as real-time state monitoring information from the SCADA system), a time series modeling method (such as ARIMA, LSTM, etc.) can be used to establish a state dynamic model of the system. The model can predict the future state evolution of the system based on the past state transition pattern and real-time data. By collecting monitoring data and corresponding state transition matrices at multiple time points, machine learning methods (such as deep learning models) can be used for training. The time series model needs to consider the time delay of state transition, external disturbances (such as power demand fluctuations) and the impact of system failures to achieve more accurate predictions. Assume that the state of the power system is modeled using an LSTM model, and the input data includes a state transition matrix and historical monitoring data, such as the output power and load of the generator. By training the model, the state evolution of each device in the system over a period of time in the future can be predicted, potential faults can be discovered in time, and maintenance or adjustments can be performed in advance.

[0050] Optionally, step S2 specifically includes: Step S21: acquiring performance index data of a large current process control system, and performing system demand statistics on the performance index data of the large current process control system, thereby obtaining demand data of the large current process control system; In this embodiment, performance index data related to system operation is collected from the large power process control system. Specifically, data such as system load, equipment operating efficiency, response time, processing capacity, failure rate, etc. can be obtained through real-time monitoring equipment or collectors. Data acquisition equipment can be implemented using tools such as PLC (programmable logic controller) and SCADA (supervisory control and data acquisition system). The collected raw data needs to be cleaned, denoised and standardized to ensure its quality and reliability. Then, these data are sorted out through data statistical analysis methods (such as regression analysis and time series analysis), and indicators such as average value, maximum value, minimum value, fluctuation range, standard deviation, etc. are calculated, and the performance requirements of the system (such as load fluctuation range, response speed, etc.) are further analyzed to extract key indicators that affect system performance, and finally form system demand data. These demand data will provide a basis for the subsequent creation of dynamic models. Assume that a power process control system monitors the operating status and load data of 10 generator sets in real time through a SCADA system. The performance indicators of each generator set include operating power, fuel consumption, temperature, pressure, etc. After data cleaning and denoising, the time series analysis method is used to calculate the average load fluctuation range, fault frequency and other data of each unit, and finally generate the demand data of the power system, such as "the maximum value of the system load demand is 400MW, the minimum value is 100MW, and the target load fluctuation range is ±10%." Step S22: mapping the program control system requirements to the system state dynamic model according to the large electric program control system requirement data, thereby obtaining the system state constraint dynamic model; In this embodiment, the system demand data will be used to map the demand of the existing system state dynamic model. The system state dynamic model is usually based on mathematical modeling of the power process, such as based on the state space model (State Space Model), Markov chain model, etc. to describe the change of system state over time. On this basis, according to the obtained system demand data, the parameters or constraints in the model are dynamically adjusted to reflect the current system demand. For example, if the system load demand increases, the relationship between the power output and the load in the model needs to be adjusted to ensure that the model can stably predict the system state in actual operation. Assume that in the power control system, the power load demand fluctuates greatly. According to the demand data analysis, the system load demand range is 100MW to 400MW. According to these demand data, the dynamic model of the system is modified, and the constraints when the load fluctuates are added to ensure the stability and safety of the power system under different load conditions. For example, a constraint is added: "When the load demand fluctuation exceeds ±10%, the system dispatching model should adjust the output of the generator set to ensure system stability." Step S23: acquiring the real-time monitoring data of the large-scale electric process control system, and performing real-time system state abnormality detection on the real-time monitoring data of the large-scale electric process control system according to the system state constraint dynamic model, thereby obtaining system abnormal state data; In this embodiment, it is necessary to obtain the monitoring data of the large power process control system in real time. These data usually come from on-site sensors, monitoring equipment or real-time data acquisition systems. The monitoring data includes data such as temperature, pressure, power, frequency, etc. of the generator set. By comparing with the system state constraint dynamic model, it is detected whether the system state is abnormal. Anomaly detection can identify data that does not conform to the normal operating state by setting a predetermined threshold or using a machine learning algorithm (such as anomaly detection algorithm, cluster analysis). Once the system detects an abnormality, it can be identified as system abnormal state data. Assume that the real-time monitoring data of the power system includes the temperature of generator set A (normal range is 70°C to 90°C), pressure (normal range is 2MPa to 4MPa), etc. If the monitoring data shows that the temperature of generator set A suddenly rises to 95°C, exceeding the set temperature threshold of 90°C, the system will mark this state as an abnormal state through the anomaly detection algorithm, trigger an alarm and record the abnormal state data.

[0051] Step S24: integrating the system operation process structure of the system state dynamic model, thereby obtaining a system operation process structure model; In this embodiment, the actual operation process topology structure obtained in the system state dynamic model is integrated to form the system operation process structure model. This process structure model not only includes the state changes of each component or link of the system, but also involves specific operation steps, control logic and decision rules. For example, in the power process control system, the system's control mode, equipment start and stop sequence, scheduling strategy, emergency response mechanism, etc. can be represented in the form of flow charts, state machine diagrams, etc. In the power system, it is assumed that power production includes multiple operation processes such as generator start and stop, power dispatching, and load distribution. Through the analysis of the system state dynamic model, these operation processes are integrated step by step to form a comprehensive operation process model. The model includes control processes such as "activate the standby generator set when the load demand increases" and "trigger the emergency shutdown operation when the system state is abnormal".

[0052] Step S25: Perform workflow environment upsampling according to the system abnormal state data and the system operation process structure model, so as to obtain scenario simulation data.

[0053] In this embodiment, the system abnormal state data and the operation process structure model are combined to simulate and upsample the workflow environment. The purpose of this step is to simulate the performance and response mechanism of the system under different abnormal states through virtual simulation or numerical simulation, and generate scenario simulation data. The sampling process can be completed through simulation experiments or virtual simulation platforms based on the dynamic response characteristics and historical data of the model. In the power system, when the system detects a fault in generator set A, the workflow simulation is performed based on the obtained abnormal state data and the operation process model. For example, it simulates how to adjust the load distribution, start the standby unit, and other operations after the fault occurs. Through multiple simulations, the state change data that may occur in the system under different circumstances is generated, and finally the scenario simulation data of "the system load balancing time is 10 minutes after the fault occurs" is formed.

[0054] Optionally, step S23 is specifically: Step S231: acquiring real-time monitoring data of a large electric process control system, and performing data preprocessing on the real-time monitoring data of the large electric process control system, thereby obtaining real-time monitoring data of the system to be analyzed; In this embodiment, various types of monitoring data are collected from the large power process control system in real time by means of Internet of Things devices, sensors, etc. These monitoring data include but are not limited to information such as current, voltage, temperature, pressure, flow, and equipment operation status. This data is sent to the central data processing system through the data acquisition gateway. The acquired data often contains noise, missing values, outliers, etc., and needs to be preprocessed. The preprocessing process includes data denoising (such as using filtering algorithms to remove high-frequency noise), data filling (filling missing values ​​through interpolation methods), and data standardization (standardizing data of different dimensions to ensure the accuracy of subsequent analysis). For example, Gaussian filtering is used to denoise temperature data, and linear interpolation is used to fill missing data for flow data, and finally the real-time monitoring data of the system to be analyzed for subsequent analysis is obtained.

[0055] Step S232: predicting the real-time dynamic state of the system to be analyzed based on the system state constraint dynamic model, thereby obtaining the real-time dynamic state data of the system; integrating the performance indicator data of the large electric process control system based on the system state constraint dynamic model, thereby obtaining the expected dynamic state data of the system; In this embodiment, based on historical data and real-time monitoring data, a system state constraint dynamic model (such as a prediction algorithm based on a Kalman filter or a deep learning model) is used to predict dynamic states. The model predicts the dynamic changes of the system in the future period of time through known real-time monitoring data, such as current, pressure, etc., combined with the state constraints of the system (such as the maximum allowable load of the equipment, temperature range, etc.). For example, a machine learning regression model is used to predict the operating state of the equipment in the next few minutes based on historical current and pressure data. Based on the performance indicator data of the system, such as power utilization, efficiency, energy consumption, etc., combined with the operating constraints of the system reflected by the system state constraint dynamic model (such as energy consumption limit, load balancing requirements, etc.), the expected state is integrated. A multivariable state integration method (such as a weighted average method, a state fusion method, etc.) is used to integrate performance indicator data from different sources to obtain the expected dynamic state data of the system under ideal conditions. For example, the power and load information of different devices are integrated into the expected dynamic state of the system as a whole using the weighted average method.

[0056] Step S233: calculating the system expected dynamic state deviation amount for the system real-time dynamic state data and the system expected dynamic state data, thereby obtaining the system dynamic state deviation amount data; In this embodiment, the deviation between the real-time dynamic state of the system and the expected dynamic state is calculated. The deviation can be obtained by calculating the difference or proportional difference between the two state data. Specifically, the real-time dynamic state data is compared with the expected dynamic state data item by item, and the deviation of each data item is calculated. For example, if the real-time temperature data is 95°C and the expected temperature is 90°C, the deviation is +5°C; if the real-time pressure data is 1.2MPa and the expected pressure is 1.0MPa, the deviation is +0.2MPa. The deviations of all data items will form a system dynamic state deviation vector, which will reflect the difference between the actual operating state of the system and the expected state. By performing statistical analysis on these deviations (such as mean and variance analysis), the operating deviation trend of the system is further analyzed.

[0057] Step S234: performing nonlinear deviation state identification on the system real-time dynamic state data according to the system dynamic state deviation data, thereby obtaining system abnormal state data.

[0058] In this embodiment, based on the calculated dynamic state deviation data, a nonlinear deviation state recognition algorithm is used to diagnose whether the system is abnormal. Common nonlinear deviation state recognition methods include classification methods based on support vector machines (SVM), anomaly detection methods based on neural networks, etc. Specifically, the system dynamic state deviation data is used as an input feature, and the data is classified by a trained model (such as an SVM model, a deep neural network, etc.). By comparing real-time data with a preset normal state model, the abnormal state is identified. For example, if the system deviation exceeds the set threshold range, the model will be marked as an abnormal state, which will trigger an alarm mechanism and notify the operation and maintenance personnel to handle it. In practical applications, an abnormal threshold can be set, for example, a deviation exceeding ±10% can be defined as an abnormal state.

[0059] Optionally, step S25 is specifically: Step S251: generating an abnormal operating system working scenario for the system operation process structure model based on the system abnormal state data, thereby obtaining an abnormal workflow scenario data set; generating a normal operating system working scenario for the system operation process structure model based on the system expected dynamic state data, thereby obtaining a normal workflow scenario data set; In this embodiment, it is necessary to obtain the abnormal state data and normal state data of the system. The abnormal state data includes data of abnormal operation such as system failure, misconfiguration, hardware failure, etc., while the normal state data includes standard parameters and behaviors of the system when it is running under expected conditions. Using the system operation process structure model (such as UML activity diagram, workflow model, etc.), these data can be mapped to the actual work scene. In the specific implementation, methods such as fault tree analysis (FTA) can be used to generate corresponding abnormal workflow scenarios according to different abnormal state data. These scenarios include the workflow of the system when a certain failure occurs, such as network delay, hard disk failure, service interruption, etc. The generation of normal work process scenarios is based on the normal dynamic behavior of the system when there is no failure or error, and is usually generated in combination with the historical operation data, expected business process and workload prediction of the system. Assume that the operation process of a production system includes steps such as "equipment startup", "temperature detection", and "data collection". The abnormal state data of the system includes equipment failure, overload, or power outage. In this step, the system uses historical fault records and real-time monitoring data to simulate abnormal conditions in the operation process by modeling abnormal state scenarios (for example, power outage or automatic equipment shutdown). This can be done by introducing conditional judgment models, such as fault tree analysis (FTA) or Markov chain models, to simulate the operating behavior of the system under specific abnormal conditions, thereby generating an abnormal workflow scenario data set. On the other hand, normal workflow scenarios are generated based on normal operating states (such as equipment operating under normal working loads) and expected dynamic state data (such as equipment load, temperature, speed, etc.). This can be done by summarizing the normal operating modes in historical data, and then simulating various workflows of the system under normal conditions through the state space model in control theory. For example, when an automated assembly line is operating normally, sensor data will show that temperature, humidity, etc. are within the ideal range, and the system will execute each operation task step by step according to the preset time step.

[0060] Step S252: performing Monte Carlo upsampling processing on the abnormal workflow scenario data set and the normal workflow scenario data set, thereby obtaining a random workflow scenario data set; In this embodiment, the Monte Carlo upsampling technique is used to expand the abnormal workflow scenario data set and the normal workflow scenario data set. The Monte Carlo method is a statistical simulation method based on random sampling. By randomly selecting data points in the abnormal and normal data sets and performing repeated sampling, a large number of different workflow scenarios are generated. Specifically, assuming that the probability of occurrence of certain operation processes (such as task scheduling, data processing, user requests, etc.) in a specific scenario is known, these scenarios can be randomly simulated through Monte Carlo sampling, and possible workflow scenarios can be generated based on the probability distribution. For example, if the probability of an operation process occurring is 0.8, random sampling is used to determine whether the operation process will occur in each scenario, and the random workflow scenario data set finally generated has high diversity and representativeness.

[0061] Step S253: parameterizing the work scenario based on the system operation process structure model, thereby obtaining a work scenario parameter set; The parameterization of the work scene in this embodiment is the process of quantifying the relevant parameters in different workflow scenes. Based on the system operation process structure model, each workflow step, task, event, etc. is abstracted into a set of parameters, such as execution time, resource consumption, response time, network bandwidth, system load, etc. In specific implementation, the system operation process can be modeled, and each node (operation step) and edge (relationship between tasks) can be associated with corresponding performance indicators, environmental variables, etc. For example, assuming that the task corresponding to a certain workflow step requires a certain CPU time, then its parameterization result will be the execution time of the task and the required CPU resources. Similarly, each workflow step is converted into specific parameters, and these parameter sets constitute the parameter set of the work scene. Each work scene is mapped into a parameter set, which includes: the execution order of the operation steps, the input and output conditions of each step, time delay, device status, environmental conditions, etc. For example, assuming that a typical operation process involves three steps of "start", "run", and "stop", the parameter set for the "start" step includes the time required for the device to start, the energy required during the startup process, etc.; while the parameter set for the "stop" step includes the time when the device stops, the temperature and pressure of the device when the stop is executed, and other data. Assume that in the assembly line operation, the startup time is usually 5 seconds and the equipment temperature is 20°C under normal environment. In this step, the state information of each link is extracted through the system model and converted into quantifiable parameters, such as "equipment startup time = 5 seconds", "ambient temperature = 20°C", etc.

[0062] Step S254: Perform workflow scenario simulation on the random workflow scenario data set and the workflow scenario parameter set to obtain scenario simulation data.

[0063] In this embodiment, the workflow scenario simulation is performed by inputting a random workflow scenario data set and a work scenario parameter set into the simulation system for simulation to generate possible workflow execution results. Through simulation, the behavior, performance and stability of the system under different work scenarios can be evaluated. Specifically, a method based on discrete event simulation (DES) can be used to perform event-driven simulation on each workflow scenario. For example, if a certain abnormal scenario is caused by data transmission delay due to insufficient network bandwidth, the impact of the scenario under different network conditions can be simulated to analyze the impact of network delay on task execution time. Through this simulation, the bottleneck or failure mode that may occur in the system under different conditions can be predicted, and the performance of the system can be optimized or emergency handling strategies can be formulated based on this. During the simulation process, simulation tools such as Simulink and Arena can be used to implement it, and the simulation results can include key performance indicators such as the completion time of each workflow, the consumption of system resources, and the response time. Assume that the system is an automated production line containing multiple work steps (such as "prepare materials", "operate equipment", and "check quality"). The generated random workflow scenario data set and parameter set are input into the simulation tool to simulate the performance of the production line under various normal and abnormal conditions. Through simulation, the system's work efficiency, reliability, failure rate and other indicators in different scenarios are evaluated. For example, in a normal operation scenario, the system completes a task within 10 seconds; in an abnormal scenario, due to equipment failure, the task completion time may be extended or even cause the task to fail.

[0064] Optionally, step S3 specifically includes: Step S31: extracting historical working scene features according to the program control system monitoring data to be analyzed, thereby obtaining historical working scene data; In this embodiment, historical monitoring data (e.g., sensor data, event logs, alarm records, etc.) are extracted from the program control system. Then, based on these data, different work scene features are identified through time series analysis and data mining technology, such as load status, equipment operation status, external environmental factors, etc. Feature selection algorithms (such as LASSO regression) can be used to screen the most representative features. Assuming that a program control system involves temperature sensor and pressure sensor data, the historical data of these sensors can be divided into "high temperature and high pressure state" and "normal state" to extract relevant work scene features.

[0065] Step S32: based on the system state dynamic model, the system historical state backtracking is performed on the program-controlled system log to be analyzed, so as to obtain historical system state data; In this embodiment, the system log to be analyzed is traced back through a system state dynamic model (such as a Markov chain or a state transition diagram model). Assuming that the program control system contains multiple subsystems (such as a temperature control system, a power system, a control system, etc.), by analyzing the historical logs, the path of historical state changes can be traced back according to the order of occurrence and timestamps of events. For example, the "power on" operation log record of a control system can be mapped to the system state conversion process through a dynamic model, and traced back to the initial startup state of the system and subsequent state changes. This process will be combined with the state transition rules in the model to perform backtracking to obtain the system state data at each time point.

[0066] Step S33: performing correlation analysis on the historical working scene data and the historical system status data, thereby obtaining scene-state correlation data; In this embodiment, statistical methods (such as Pearson correlation coefficient, mutual information method or regression analysis) are used to analyze the correlation between historical work scene data and system status data. For example, by analyzing the changing patterns of system status (such as "power demand" and "temperature change") under specific work scenes (such as "high-load working state"), the behavior pattern of the system and its correlation with the external environment can be revealed. Assuming that in the "high-load" working scene, the system state shows a correlation between high temperature and high pressure, it indicates that under high-load conditions, there is a strong correlation between temperature and pressure, reflecting the state characteristics of the system under high load.

[0067] Step S34: simulating the state of the program-controlled system on the system state dynamic model according to the scene-state correlation data and using the scene simulation data, thereby obtaining the program-controlled system simulation state data; In this embodiment, based on the obtained scene-state correlation data, in this step, the system state dynamic model is simulated in combination with the scene simulation data. For example, a multivariate regression model based on historical data and scene-state correlation data is used to construct a system state simulation model, and the system state of the system state dynamic model is predicted by the scene simulation data (such as workload, environmental changes and other factors). Assuming that for a temperature control system, the scene data includes an operating temperature range, the system state dynamic model can simulate and predict the temperature change trend under different environmental conditions based on the data. This simulation result can help optimize the system operation strategy or provide early warning of potential failures.

[0068] Step S35: Perform simulation state pattern recognition based on the program control system simulation state data, thereby obtaining system process simulation state pattern data.

[0069] In this embodiment, the simulation state data of the program control system is analyzed by pattern recognition algorithms (such as K-means clustering, support vector machine SVM, neural network, etc.) to identify different simulation state modes. For example, assuming that the simulation state data includes multiple indicators (such as pressure, temperature, flow, etc.), through cluster analysis, the system can divide the simulation state data into several modes, such as "normal operation mode", "overload mode", "fault warning mode", etc. This process can use unsupervised learning algorithms (such as K-means) to cluster the simulation state data, identify the patterns related to different states that appear in actual operation, and provide support for subsequent process optimization and early warning system design.

[0070] Optionally, step S35 is specifically: Step S351: setting the system state mode category based on the large current process control system performance indicator data, thereby obtaining the system state mode category data; In this embodiment, the system state mode category is set according to the performance indicators of the equipment (such as temperature, pressure, flow, current, etc.). Different state modes are defined by analyzing historical performance indicator data and combining the normal operating state and abnormal state of the equipment. For example, when the temperature and pressure exceed the preset threshold at the same time, it can be defined as an "overload" state; when the flow is normal but the temperature is low, it is defined as an "inefficient" state. The data is classified by setting a threshold or applying a classification method based on machine learning (such as a decision tree, support vector machine, etc.) to obtain the state mode category data of the system. Assuming that the control system of a chemical plant is monitored, the system will continuously record the data of various sensors. Use data processing tools (such as Python's Pandas or Scikit-learn) to read these data and apply the threshold method (for example, the temperature exceeds 100°C as the "overheating" mode) to classify the data, and finally form the state mode category data.

[0071] Step S352: extracting simulation state features from the program control system simulation state data, thereby obtaining simulation state feature data; In this embodiment, after obtaining the simulation state data of the program control system, it is necessary to extract representative feature information from it. Common features include mean value, standard deviation, maximum value, minimum value and data trend (such as data change rate or periodicity). Assuming that the simulation data includes equipment temperature, flow rate and pressure, the feature extraction process can use time series analysis methods, such as the sliding window method to calculate the average temperature and pressure fluctuation amplitude of each time period, so as to extract features such as "temperature fluctuation" or "pressure stability". For example, for pressure data, the simulation state data is first preprocessed (noise removal, standardization, etc.), and then the window function (such as sliding window) is used to extract the average value and change rate in each time period to form state feature data, which can be used as input for the next step of simulation state similarity calculation.

[0072] Step S353: performing simulation state similarity calculation according to the simulation state feature data, thereby obtaining simulation state feature similarity data; In this embodiment, the simulation state similarity calculation is usually based on a certain similarity measurement method, such as Euclidean distance, cosine similarity or Manhattan distance. For each simulation state, the similarity data between the simulation states is obtained by calculating the similarity between its feature vector and the feature vectors of other states. For example, assuming that there are two simulation states, their features are (temperature, pressure) = (100°C, 50MPa) and (102°C, 52MPa), respectively. The Euclidean distance formula can be used to calculate the similarity between the two to obtain similarity data. Based on the feature data of the simulation state, the similarity between the features of each two simulation states is calculated. Assuming that the extracted state feature data contains multiple dimensions (such as temperature, pressure, flow), the Euclidean distance formula is used for similarity calculation, and then the similarity matrix between each two simulation states is obtained as the basis for the next step of clustering calculation.

[0073] Step S354: performing simulation state clustering calculation on the program control system simulation state data based on the simulation state feature similarity data, thereby obtaining the program control system clustered simulation state data; In this embodiment, clustering calculation is performed on the basis of the simulation state similarity data. Common clustering methods include K-means clustering, DBSCAN (density clustering) and hierarchical clustering. Through clustering, similar simulation states are divided into the same category to form a clustering result. For example, in K-means clustering, a preset K value (such as 3) is first selected, and then the category to which each state belongs is iteratively optimized so that the feature difference of the state within each class is as small as possible, and the difference between the states of the classes is as large as possible. The final result is that each simulation state is classified into a specific state mode category. Taking K-means clustering as an example, the similarity matrix calculated previously is used as input data, and K=3 is first selected as the number of clusters. Then, the simulation state data is clustered by the K-means algorithm, and finally three different simulation state clusters (such as "normal state", "minor fault state" and "serious fault state") are obtained.

[0074] Step S355: Perform simulation state pattern recognition on the clustered simulation state data of the program-controlled system according to the system state pattern category data, thereby obtaining system process simulation state pattern data.

[0075] In this embodiment, based on the set system state mode category data, the clustering results are pattern recognized to determine the actual meaning of each cluster cluster. This process can be completed by comparing the actual operating performance of each type of simulation state with the corresponding mode category. For example, the state in the "normal state" cluster is compared with the state in the "overload" cluster, and the relationship between the clustering results and the preset mode category is analyzed, and finally a specific process mode (such as "normal operation", "high temperature overload", etc.) is assigned to each clustering result. Assume that in step S354, three types of clustering results are obtained, namely cluster 1, cluster 2 and cluster 3. By comparing the performance of normal operation state, minor fault state and major fault state in historical data, cluster 1 is matched with the "normal operation" state mode, cluster 2 is matched with the "minor fault" state, and cluster 3 is matched with the "major fault" state. In this way, the recognition of the simulation state mode is completed, and the "system process simulation state mode data" is finally generated.

[0076] Optionally, the present specification further provides an automated testing system applicable to large current program control, which is used to execute the automated testing method applicable to large current program control as described above, and the automated testing system applicable to large current program control includes: The system state modeling module is used to obtain the log of the large-scale electric process control system and the monitoring data of the large-scale electric process control system, and to model the workflow topology structure according to the log of the large-scale electric process control system, so as to obtain the workflow topology structure model; and to model the system state according to the workflow topology structure model and the monitoring data of the large-scale electric process control system, so as to obtain the system state dynamic model; The system state anomaly detection module is used to map the system state dynamic model to the requirements of the program-controlled system, thereby obtaining the system state constraint dynamic model, and to perform real-time system state anomaly detection based on the system state constraint dynamic model, thereby obtaining system abnormal state data; and to perform workflow environment upsampling based on the system abnormal state data and the system state dynamic model, thereby obtaining scenario simulation data; The program-controlled system state simulation module is used to simulate the program-controlled system state on the system state dynamic model according to the scenario simulation data, thereby obtaining the program-controlled system simulation state data, and to perform simulation state pattern recognition based on the program-controlled system simulation state data, thereby obtaining the system process simulation state pattern data; A mode score calculation module is used to calculate the system state mode score of the system process simulation state mode data based on the system state constraint dynamic model, so as to obtain the program control system state score data; The repair strategy selection module is used to select the process control system repair strategy based on the program control system status score data and the large-scale program control system log, so as to obtain the system repair strategy and upload it to the large-scale program control system to execute the program repair task.

[0077] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0078] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. An automated testing method suitable for large current process control, characterized in that: The following steps are involved: Step S1: Obtain the log of the large electric process control system and the monitoring data of the large electric process control system, and perform workflow topology modeling according to the log of the large electric process control system, so as to obtain the workflow topology model; perform system state modeling according to the workflow topology model and the monitoring data of the large electric process control system, so as to obtain the system state dynamic model; Step S2: mapping the system state dynamic model to the program control system requirements, thereby obtaining the system state constraint dynamic model, and performing real-time system state anomaly detection according to the system state constraint dynamic model, thereby obtaining system abnormal state data; Perform workflow environment upsampling based on system abnormal state data and system state dynamic model to obtain scenario simulation data; Step S3: performing program-controlled system state simulation on the system state dynamic model according to the scenario simulation data, thereby obtaining program-controlled system simulation state data, and performing simulation state pattern recognition based on the program-controlled system simulation state data, thereby obtaining system process simulation state pattern data; Step S4: Calculate the system state mode score of the system process simulation state mode data based on the system state constraint dynamic model, thereby obtaining the program control system state score data; Step S5: Select a process control system repair strategy based on the program control system status score data and the large-scale program control system log, thereby obtaining a system repair strategy, and upload it to the large-scale program control system to execute the program repair task.

2. The automated testing method for large current process control according to claim 1, characterized in that: Step S1 is specifically as follows: Step S11: Obtaining the large current process control system log and the large current process control system monitoring data; Step S12: performing data preprocessing on the large electric process control system log and the large electric process control system monitoring data respectively, so as to obtain the process control system log to be analyzed and the process control system monitoring data to be analyzed; Step S13: integrating system function module features according to the program-controlled system log to be analyzed, thereby obtaining system function module data; Step S14: Modeling the workflow topology structure based on the system function module data, thereby obtaining a workflow topology structure model; Step S15: System state modeling is performed according to the workflow topology model and the large power process control system monitoring data, so as to obtain a system state dynamic model.

3. The automated testing method for large current process control according to claim 2, characterized in that: Step S14 is specifically as follows: Step S141: extracting function module interaction events and function module communication from system function module data, thereby obtaining function module interaction data and function module communication data; Step S142: Connecting the function module communication nodes of the program control system according to the function module communication data, thereby obtaining the function module communication node connection data; Step S143: performing a function module interaction event time series analysis on the function module interaction data, thereby obtaining the function module interaction event time series data; Step S144: identifying time series dependency relationships based on the time series data of the functional module interaction events, thereby obtaining functional module dependency relationship data; Step S145: Modeling the workflow topology structure according to the functional module communication node connection data and the functional module dependency data, thereby obtaining a workflow topology structure model.

4. The automated testing method for large current process control according to claim 2, characterized in that: Step S15 is specifically as follows: Step S151: define the state space based on the workflow topology model, so as to obtain the system state space; Step S152: extracting system operation data from the large-scale process control system log, thereby obtaining system operation data, and defining state transition rules for the system operation data and the large-scale process control system monitoring data, thereby obtaining state transition rules; Step S153: performing state transfer matrix statistics according to the state transfer rule and the system state space, thereby obtaining the system state transfer matrix; Step S154: Perform time series dynamic modeling based on the system state transfer matrix and the large-scale power process control system monitoring data to obtain a system state dynamic model.

5. The automated testing method for large current process control according to claim 1, characterized in that: Step S2 is specifically as follows: Step S21: acquiring performance index data of a large current process control system, and performing system demand statistics on the performance index data of the large current process control system, thereby obtaining demand data of the large current process control system; Step S22: mapping the program control system requirements to the system state dynamic model according to the large electric program control system requirement data, thereby obtaining the system state constraint dynamic model; Step S23: acquiring the real-time monitoring data of the large-scale electric process control system, and performing real-time system state abnormality detection on the real-time monitoring data of the large-scale electric process control system according to the system state constraint dynamic model, thereby obtaining system abnormal state data; Step S24: integrating the system operation process structure of the system state dynamic model, thereby obtaining a system operation process structure model; Step S25: Perform workflow environment upsampling according to the system abnormal state data and the system operation process structure model, so as to obtain scenario simulation data.

6. The automated testing method for large current process control according to claim 5, characterized in that: Step S23 is specifically as follows: Step S231: acquiring real-time monitoring data of a large electric process control system, and performing data preprocessing on the real-time monitoring data of the large electric process control system, thereby obtaining real-time monitoring data of the system to be analyzed; Step S232: predicting the real-time dynamic state of the system to be analyzed based on the system state constraint dynamic model, thereby obtaining the real-time dynamic state data of the system; integrating the performance indicator data of the large electric process control system based on the system state constraint dynamic model, thereby obtaining the expected dynamic state data of the system; Step S233: calculating the system expected dynamic state deviation amount for the system real-time dynamic state data and the system expected dynamic state data, thereby obtaining the system dynamic state deviation amount data; Step S234: performing nonlinear deviation state identification on the system real-time dynamic state data according to the system dynamic state deviation data, thereby obtaining system abnormal state data.

7. The automated testing method for large current process control according to claim 5, characterized in that: Step S25 is specifically as follows: Step S251: generating an abnormal operating system working scenario for the system operation process structure model based on the system abnormal state data, thereby obtaining an abnormal workflow scenario data set; generating a normal operating system working scenario for the system operation process structure model based on the system expected dynamic state data, thereby obtaining a normal workflow scenario data set; Step S252: performing Monte Carlo upsampling processing on the abnormal workflow scenario data set and the normal workflow scenario data set, thereby obtaining a random workflow scenario data set; Step S253: parameterizing the work scenario based on the system operation process structure model, thereby obtaining a work scenario parameter set; Step S254: Perform workflow scenario simulation on the random workflow scenario data set and the workflow scenario parameter set to obtain scenario simulation data.

8. The automated testing method for large current process control according to claim 1, characterized in that: Step S3 is specifically as follows: Step S31: extracting historical working scene features according to the program control system monitoring data to be analyzed, thereby obtaining historical working scene data; Step S32: based on the system state dynamic model, the system historical state backtracking is performed on the program-controlled system log to be analyzed, so as to obtain historical system state data; Step S33: performing correlation analysis on the historical working scene data and the historical system status data, thereby obtaining scene-status correlation data; Step S34: simulating the state of the program-controlled system on the system state dynamic model according to the scene-state correlation data and using the scene simulation data, thereby obtaining the program-controlled system simulation state data; Step S35: Perform simulation state pattern recognition based on the program control system simulation state data, thereby obtaining system process simulation state pattern data.

9. The automated testing method for large current process control according to claim 8, characterized in that: Step S35 is specifically as follows: Step S351: setting the system state mode category based on the large current process control system performance indicator data, thereby obtaining the system state mode category data; Step S352: extracting simulation state features from the program control system simulation state data, thereby obtaining simulation state feature data; Step S353: performing simulation state similarity calculation according to the simulation state feature data, thereby obtaining simulation state feature similarity data; Step S354: performing simulation state clustering calculation on the program control system simulation state data based on the simulation state feature similarity data, thereby obtaining the program control system clustered simulation state data; Step S355: Perform simulation state pattern recognition on the clustered simulation state data of the program-controlled system according to the system state pattern category data, thereby obtaining system process simulation state pattern data.

10. An automated testing system suitable for large current process control, characterized in that: Used to execute the automated testing method applicable to large current process control as claimed in claim 1, the automated testing system applicable to large current process control comprises: The system state modeling module is used to obtain the log of the large-scale electric process control system and the monitoring data of the large-scale electric process control system, and to model the workflow topology structure according to the log of the large-scale electric process control system, so as to obtain the workflow topology structure model; and to model the system state according to the workflow topology structure model and the monitoring data of the large-scale electric process control system, so as to obtain the system state dynamic model; The system state anomaly detection module is used to map the system state dynamic model to the requirements of the program-controlled system, thereby obtaining the system state constraint dynamic model, and to perform real-time system state anomaly detection based on the system state constraint dynamic model, thereby obtaining system abnormal state data; and to perform workflow environment upsampling based on the system abnormal state data and the system state dynamic model, thereby obtaining scenario simulation data; The program-controlled system state simulation module is used to simulate the program-controlled system state on the system state dynamic model according to the scenario simulation data, thereby obtaining the program-controlled system simulation state data, and to perform simulation state pattern recognition based on the program-controlled system simulation state data, thereby obtaining the system process simulation state pattern data; A mode score calculation module is used to calculate the system state mode score of the system process simulation state mode data based on the system state constraint dynamic model, so as to obtain the program control system state score data; The repair strategy selection module is used to select the process control system repair strategy based on the program control system status score data and the large-scale program control system log, so as to obtain the system repair strategy and upload it to the large-scale program control system to execute the program repair task.

Images