A typical fault diagnosis system and method for nuclear power plants

Abstract

The present application relates to the field of nuclear power operation, and particularly relates to a nuclear power plant typical fault diagnosis system and method.The system comprises a data processing module for accessing, processing, storing and querying unit operation data; a logic analysis module for automatically analyzing and operating interfaces and internal operation blocks in unit DCS configuration files; a knowledge model module for quantitatively expressing fault types, features, causal relationships and weight relationships between fault types and features; an algorithm module comprising a time series data analysis algorithm and a knowledge reasoning algorithm; the time series data analysis algorithm is used for tracing the first signal of an event, and the knowledge reasoning algorithm is used for locating an event cause or a fault point; a man-machine interaction module for displaying analysis results, configuring algorithms, editing knowledge models and managing personnel and permissions.The present application realizes automatic analysis of fault causes and improves the efficiency and accuracy of event cause analysis and positioning.

Description

Technical Field

[0001] This invention relates to the field of nuclear power plant operation, and in particular to a typical fault diagnosis system and method for nuclear power plants. Background Technology

[0002] Incident analysis is a crucial component of ensuring nuclear safety. Currently, power plants face significant challenges in incident analysis. Not only are data collection, parameter screening, and trend analysis cumbersome and time-consuming, lacking dedicated tools and resulting in low efficiency, but the analysis of incident causes also relies heavily on maintenance personnel's knowledge and experience, involving detailed data analysis and comparison. This reliance is highly dependent on the skill level of the personnel involved. By leveraging digital and intelligent technologies, fully utilizing power plant operational data, and digitizing and expressing rich power plant operational knowledge and experience to create digital intelligent analysis tools, effective technical support can be provided for power plant incident analysis activities, improving efficiency and quality.

[0003] Currently, when nuclear power plants analyze the causes of abnormal events such as unexpected shutdowns, reactor shutdowns, special equipment actions, critical equipment protection actions, and comprehensive alarms, they generally use data analysis methods provided by the DCS system (including alarm information, historical parameter trends, etc.) to analyze and locate the causes of the events. On the one hand, this requires a high level of knowledge, experience, and skills from the operators, and on the other hand, it consumes a lot of time and energy. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a typical fault diagnosis system and method for nuclear power plants, so as to realize the automatic analysis of fault causes and improve the efficiency and accuracy of event cause analysis and location.

[0005] This invention provides a nuclear power plant operation event-assisted intelligent analysis system based on logical parsing and knowledge models, comprising:

[0006] The data processing module is used for accessing, processing, storing, and querying unit operation data;

[0007] The logic parsing module is used for automatic parsing and calculation of interfaces and internal calculation blocks in the unit's DCS configuration file;

[0008] The knowledge model module is used to quantitatively express fault types, features, and the causal and weighted relationships between fault types and features;

[0009] The algorithm module includes a time-series data analysis algorithm and a knowledge reasoning algorithm; the time-series data analysis algorithm is used to trace the initial signal of an event, and the knowledge reasoning algorithm is used to locate the cause of the event or the fault point.

[0010] The human-computer interaction module is used to display analysis results, configure algorithms, edit knowledge models, and manage personnel and permissions.

[0011] In one specific embodiment of the present invention, the data processing module includes a time-series database, a data communication interface module, a data preprocessing module, and a data management module;

[0012] The time-series database is used to store unit operation data;

[0013] The data communication interface module is used to transfer data from the unit's DCS operating database system to this system.

[0014] The data preprocessing module is used to preprocess the raw data;

[0015] The data management module is used to transmit the processed data to the algorithm module.

[0016] In one specific embodiment of the present invention, the preprocessing includes data incompleteness, removal of singular values, noise reduction, or smoothing.

[0017] In one specific embodiment of the present invention, the logic parsing module includes a logic parsing tool and a runtime environment for the parsed logic program;

[0018] The logic parsing tool is used to parse various operation blocks in the original configuration logic.

[0019] In one specific embodiment of the present invention, the knowledge model module includes a knowledge editing tool and an event knowledge model.

[0020] The knowledge editing tool is used to build an event knowledge model, specifically including defining and quantifying the hypothetical fault types related to various events, the corresponding characteristics of each fault type, the causal relationship and weight relationship between the two;

[0021] The event knowledge model is compiled to generate an executable program.

[0022] In one specific embodiment of the present invention, the human-computer interaction module displays the monitoring status of various operational events in real time. When an event is detected, the background time-series data analysis algorithm is automatically started, and the operational data within a certain period before the event occurs is automatically extracted as input to begin tracing and locating the initial signal of the event. After the initial signal is identified, the relevant logic diagram and source sensor information of the signal are displayed. At the same time, the knowledge reasoning algorithm is started to analyze and diagnose the cause of the event, and finally displays a list of possible faults.

[0023] This invention provides a typical fault diagnosis method for nuclear power plants, comprising the following steps:

[0024] Step 1: Analyze the operational knowledge and experience related to the causes of nuclear power plant incidents and typical failures, and establish an event knowledge model based on the analysis;

[0025] Step 2: Receive the DCS configuration file of the operating unit, perform logical parsing and calculation on it, and obtain the logical parsing result;

[0026] Step 3: Based on the logical parsing results, perform reverse analysis to locate the initial signal of the event;

[0027] Step 4: Import the relevant operational data from the unit operation database into the knowledge reasoning algorithm, perform reasoning calculations, and locate the cause of the event or the fault point.

[0028] In a specific embodiment of the present invention, in step 1, the event knowledge model matches various collected fault causes, feature parameters, and causal relationship probability matrices.

[0029] The event knowledge model is continuously revised and optimized using simulation and real sample data.

[0030] In a specific embodiment of the present invention, step 3 specifically includes:

[0031] Real-time scanning and monitoring of the trigger status of various events.

[0032] When an event is detected, the tracing mechanism is automatically activated. It starts to retrieve unit operation data from the time series database, traverses all event-related signal values ​​and performs logical operations to calculate the relevant signals, and obtains the signal point that flips first, which is the first signal of the event. At the same time, the analysis results are stored in the time series database and pushed and displayed on the human-machine interface.

[0033] In one specific embodiment of the present invention, step 4 specifically includes:

[0034] Using unit event operation data as input, and based on the event knowledge model built by the user and its defined causal relationships and relationship weight values, the system performs inference calculations on the causes of events until faults that match the characteristic data are identified. At the same time, it provides a list of all possible faults according to the calculated probability, providing reference and guidance for the user's analysis activities.

[0035] The results of the reasoning operations are fed back to the knowledge reasoning algorithm for continuous correction and optimization.

[0036] Compared with existing technologies, the nuclear power plant typical fault diagnosis system and method of this invention, based on logical analysis and knowledge reasoning, utilizes big data for intelligent analysis, improving the efficiency and accuracy of incident cause analysis and localization. It eliminates a significant amount of manual analysis work, including reviewing historical curves of relevant parameters, operating logs, and other documents, and then combining alarm information and operational information for correlation and comparative analysis. This reduces the workload of maintenance personnel after an incident or fault occurs, allowing them to devote more time and energy to the control and stabilization of the unit's status, ensuring the safety of the nuclear power plant unit. Attached Figure Description

[0037] Figure 1 This diagram illustrates the architecture of a typical fault diagnosis system for a nuclear power plant.

[0038] Figure 2 This represents the network structure diagram of the knowledge model of this invention;

[0039] Figure 3 This represents a breakdown diagram of the event knowledge of the present invention. Detailed Implementation

[0040] To further understand the present invention, embodiments of the present invention are described below in conjunction with examples. However, it should be understood that these descriptions are only for further illustrating the features and advantages of the present invention, and not for limiting the present invention.

[0041] Embodiments of the present invention disclose a nuclear power plant operation event-assisted intelligent analysis system based on logical parsing and knowledge models, such as... Figure 1 As shown, it includes: a data processing module, a logic parsing module, a knowledge model module, an algorithm module, and a human-computer interaction module;

[0042] The data processing module is a data platform, including a time-series database, a data communication interface module, a data preprocessing module, and a data management module, which supports the access, processing, storage, and querying of unit operation data;

[0043] The data processing module can access and store data from the unit's operating database system. It can also perform a series of preprocessing steps on this raw data, including data completion, outlier removal, noise reduction, and smoothing, to ensure data quality and avoid calculation errors and frequent false alarms during event-assisted intelligent analysis.

[0044] Specifically, data from the unit's DCS operation database system is accessed and stored in the system's time-series database via a one-way data communication interface module. When tracing the initial signal of an event and locating its cause, the algorithm module reads the data through the data management module. If the original data has quality issues, it is preprocessed by the data preprocessing module before being read. The specific preprocessing methods are as follows: data completion, noise reduction, and outlier removal are performed by developing preprocessing functions such as forward interpolation. Only after processing can the data meet the usage requirements.

[0045] The data processing module supports real-time access to unit operation data, and supports data slicing, data export, point-by-point query, time-by-time query, and data trend display.

[0046] The logic parsing module includes a logic parsing tool and a runtime environment for the parsed logic program, supporting automatic parsing and calculation of interfaces and internal operation blocks in the unit's DCS configuration file;

[0047] The logic parsing module can parse and perform calculations on the DCS configuration files of the generator set.

[0048] The parsed content includes the drawing name, basic drawing information, inter-drawing link information, equipment information within the drawing, and logical operation block information within the drawing.

[0049] The configuration file can be in the format of CAD, Visio, XML, etc.

[0050] The system uses text recognition technology to parse the DCS data interface names in the configuration logic; in addition, it uses a logic parsing tool to parse various operation blocks in the original configuration logic, including AND, OR, and NOT logic operation blocks, voting calculation blocks such as 3-out-of-2 and 4-out-of-2, threshold calculation blocks, condition selection calculation blocks, delay calculation blocks, and lead-lag calculation blocks.

[0051] The parsed file is available for user access via a web interface in the human-computer interaction module, while the calculated values ​​are dynamically displayed on the diagram screen. Simultaneously, based on the logical parsing results, the time-series data analysis algorithm can calculate and determine the initial signal.

[0052] The knowledge model module is a digital representation of the knowledge and experience of nuclear power plant operation and maintenance personnel, refined and summarized according to specific rules.

[0053] The knowledge model module includes a knowledge editing tool and an event knowledge model built based on operational knowledge and experience. It supports the quantitative expression of causal relationships according to a "event-fault-feature" ternary structure, such as... Figure 3 As shown;

[0054] Knowledge editing tools are used to build event knowledge models, specifically including defining and quantifying hypothetical fault types related to various events, the corresponding characteristics of each fault type, and the causal and weighted relationships between them; the network structure of the event knowledge model is as follows: Figure 2 As shown;

[0055] The knowledge editing tool supports graphical modeling. The completed event knowledge model is compiled to generate an executable program and is then integrated with the reasoning program of the knowledge reasoning algorithm.

[0056] The status of fault characteristics is divided into normal, abnormally high, and abnormally low. In addition, it also supports the definition of gradual abnormality of characteristics.

[0057] The causal and weighted relationships between faults and characteristics are determined based on unit vertical data and the knowledge and experience of operation and maintenance personnel, and are continuously corrected and optimized through a large amount of simulation and real sample data.

[0058] The event knowledge model can be modularly constructed according to event type, and the system also supports the configuration and selection of event knowledge models to improve analysis efficiency;

[0059] The algorithm module includes a time-series data analysis algorithm and a knowledge reasoning algorithm. The time-series data analysis algorithm reads the event logic signal configuration table parsed by the logic analysis module to obtain all event-related signal points and logic operation rules. On the one hand, it performs real-time scanning and monitoring of the triggering status of various events. On the other hand, when an event is detected, it immediately and automatically starts the tracing mechanism, begins to obtain unit operation data from the time-series database, and performs backward calculations on the relevant signals by traversing all event-related signal values ​​and performing logical operations. It adds / subtracts the lead / lag time in various operation blocks to obtain the signal point that flips first, which is the first signal of the event. At the same time, the analysis results are stored in the database and pushed and displayed on the human-computer interaction interface.

[0060] For knowledge reasoning algorithms, the system takes unit event operation data as input and performs reasoning calculations on the causes of events based on the event knowledge model built by the user and its defined causal relationships and relationship weight values ​​until the faults that match the feature data are identified. At the same time, it provides a list of all possible faults according to the calculated probability, providing reference and guidance for the user's analysis activities.

[0061] The knowledge reasoning algorithm adopts the mature algorithm principle based on "Bayesian theory" and integrates general algorithm programs to support event cause analysis and diagnosis based on knowledge models and running data;

[0062] The human-computer interaction module includes modules for displaying analysis results, configuring algorithms, editing event knowledge models, and basic system management such as personnel and permissions.

[0063] The human-computer interaction module can display the monitoring status of various operational events in real time. When an event is detected, the background time-series data analysis algorithm is automatically started, and the operational data within a certain period before the event occurs is automatically extracted as input to begin tracing and locating the initial signal of the event. After the initial signal is identified, the relevant logic diagram and source sensor information of the signal can be displayed. At the same time, the knowledge reasoning algorithm is started to analyze and diagnose the cause of the event, and finally displays a list of possible faults.

[0064] Embodiments of the present invention also disclose a typical fault diagnosis method for nuclear power plants, comprising the following steps:

[0065] Step 1: Analyze the operational knowledge and experience related to the causes of nuclear power plant incidents and typical failures, and establish an event knowledge model based on the analysis;

[0066] The analysis includes faults, features, and the causal relationship and relationship weights between them;

[0067] The event knowledge model will match various collected fault causes, feature parameters, and causal relationship probability matrices.

[0068] Preferably, the causal and weighted relationships between faults and features are determined based on unit vertical data and the knowledge and experience of operation and maintenance personnel, and are continuously corrected and optimized through a large amount of simulation and real sample data;

[0069] Step 2: Receive the DCS configuration file of the operating unit, perform logical parsing and calculation on it, and obtain the logical parsing result;

[0070] Specifically, the system receives the DCS configuration file of the operating unit and forms an event logic signal configuration table; through the event logic signal configuration table, it obtains all event-related signal points and logic operation rules.

[0071] Step 3: Based on the logical parsing results, perform reverse analysis to locate the initial signal of the event; specifically including:

[0072] Real-time scanning and monitoring of the trigger status of various events.

[0073] When an event is detected, the tracing mechanism is automatically activated. It starts to retrieve unit operation data from the time series database, traverses all event-related signal values ​​and performs logical operations to calculate the relevant signals, and obtains the signal point that flips first, which is the first signal of the event. At the same time, the analysis results are stored in the time series database and pushed and displayed on the human-machine interface.

[0074] Step 4: Import the relevant operational data from the unit operation database into the knowledge reasoning algorithm, perform reasoning calculations, and locate the cause of the event or the fault point.

[0075] Specifically, it includes:

[0076] Using unit event operation data as input, and based on the event knowledge model built by the user and its defined causal relationships and relationship weights, the system performs inference calculations on the causes of events until faults matching the characteristic data are identified. At the same time, it provides a list of all possible faults according to the calculated probability, providing reference and guidance for the user's analysis activities.

[0077] It also includes feeding the results of reasoning operations back to the knowledge reasoning algorithm for continuous correction and optimization.

[0078] The innovation of this invention lies in:

[0079] 1. This invention takes nuclear power plant operation events as its starting point and, based on the nuclear power plant event analysis process and requirements, carries out the design, development, and integrated application of the system;

[0080] 2. The parsing tool for the DCS configuration file of the unit in this invention can automatically parse and calculate the DCS configuration logic and confirm the function and role of each logic block;

[0081] 3. The analysis algorithm for unit operation sequence data in this invention can perform reverse calculation of logical time and support the tracing of the first signal of various operating events;

[0082] 4. The knowledge editing tool in this invention supports the digital expression of the knowledge and experience of nuclear power plant operation and maintenance personnel;

[0083] 5. This invention establishes a knowledge model around typical operating events in nuclear power plants, which can support the analysis and diagnosis of the causes of various operating events; the process for analyzing and diagnosing the causes of nuclear power plant events based on knowledge reasoning technology is described below. Figure 3 As shown.

[0084] 6. This invention supports online and offline access to nuclear power plant operation data and can automatically perform data analysis using algorithm programs;

[0085] 7. The human-computer interaction module of this invention can monitor the event status in real time, display the logical function diagram of various event signals, and display the analysis results of the initial signal and the cause of the event.

[0086] The above description of the embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. It should be noted that those skilled in the art can make several improvements and modifications to the present invention without departing from the principles of the present invention, and these improvements and modifications also fall within the protection scope of the claims of the present invention.

[0087] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily 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 invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A nuclear power plant operation event-assisted intelligent analysis system based on logical parsing and knowledge models, characterized in that, include: The data processing module is used for accessing, processing, storing, and querying unit operation data. The data processing module includes a time-series database, a data communication interface module, a data preprocessing module, and a data management module. The time-series database stores unit operation data. The data communication interface module is used to transfer data from the unit's DCS operation database system to this system. The data preprocessing module is used to preprocess the raw data; The data management module is used to transmit the processed data to the algorithm module; The logic parsing module is used for automatic parsing and calculation of interfaces and internal calculation blocks in the unit's DCS configuration file; The knowledge model module is used to quantitatively express fault types, corresponding features of fault types, and the causal and weighted relationships between fault types and their corresponding features. The knowledge model module includes a knowledge editing tool and an event knowledge model. The knowledge editing tool is used to build an event knowledge model, specifically including defining and quantifying the hypothetical fault types related to the event, the corresponding characteristics of each fault type, and the causal and weighted relationships between the hypothetical fault types and the corresponding characteristics of the fault types; the event knowledge model is compiled to generate an executable program; The algorithm module includes time-series data analysis algorithms and knowledge reasoning algorithms; The time-series data analysis algorithm is used to trace the initial signal of an event, and the knowledge reasoning algorithm is used to locate the cause of the event or the fault point. The human-computer interaction module is used to display analysis results, configure algorithms, edit knowledge models, and manage personnel and permissions. The human-computer interaction module displays the monitoring status of various operational events in real time. When an event is detected, the background time-series data analysis algorithm is automatically started, and the operational data within a certain period before the event occurs is automatically extracted as input to begin tracing and locating the initial signal of the event. After the initial signal is identified, the relevant logic diagram and source sensor information of the signal are displayed. At the same time, the knowledge reasoning algorithm is started to analyze and diagnose the cause of the event, and finally displays a list of possible faults.

2. The nuclear power plant operation event-assisted intelligent analysis system based on logical parsing and knowledge models according to claim 1, characterized in that, The preprocessing includes handling incomplete data, removing outliers, and noise reduction or smoothing.

3. The nuclear power plant operation event-assisted intelligent analysis system based on logical parsing and knowledge models according to claim 1, characterized in that, The logic parsing module includes logic parsing tools and the runtime environment for the parsed logic program; The logic parsing tool is used to parse the operational blocks in the original configuration logic.

4. A typical fault diagnosis method for nuclear power plants, characterized in that, Includes the following steps: Step 1: Analyze the operational knowledge and experience related to the causes of nuclear power plant incidents and typical failures, and establish an event knowledge model based on the analysis; The event knowledge model will match various collected fault causes, feature parameters, and causal relationship probability matrices; The event knowledge model is continuously revised and optimized through simulation and real sample data; Step 2: Receive the DCS configuration file of the operating unit, and perform logical parsing and calculation on it to obtain the logical parsing result; Step 3: Based on the logical parsing results, perform reverse analysis to locate the initial signal of the event; Specifically, this includes: real-time scanning and monitoring of the trigger status of various events. When an event is detected, the tracing mechanism is automatically activated. It starts to retrieve unit operation data from the time series database, it iterates through all event-related signal values ​​and performs logical operations to calculate the relevant signals, and obtains the signal point that flips first, which is the first signal of the event. At the same time, the analysis results are stored in the time series database and pushed and displayed on the human-computer interaction interface. Step 4: Import the relevant operational data from the unit operation database into the knowledge reasoning algorithm, perform reasoning calculations, and locate the cause of the event or the fault point; Specifically, this includes: taking unit event operation data as input, and based on the event knowledge model built by the user and its defined causal relationships and relationship weights, performing inference calculations on the causes of events until faults matching the characteristic data are identified, and providing a list of all possible faults according to the calculated probability, providing reference and guidance for the user's analysis activities; The results of the reasoning operations are fed back to the knowledge reasoning algorithm for continuous correction and optimization.

Images