Nuclear power key equipment reliability intelligent analysis system and method

Abstract

The invention relates to a nuclear power key equipment reliability intelligent analysis system and method. The system comprises a knowledge graph construction module, a double attention mechanism neural network model, a thermodynamic diagram supervision module and a long-term prediction terminal. The knowledge graph construction module constructs an equipment disassembly tree knowledge graph and stores fault mode data; the dual attention mechanism neural network model generates a dual attention mechanism model, captures a space dimension and a time dimension, constructs a neural network model and a real-time prediction model, and synchronously updates and optimizes the models; the thermodynamic diagram supervision module dynamically updates a target three-dimensional thermodynamic diagram, supervises a dynamic reliability thermodynamic diagram engine, and carries out visual supervision processing on fault propagation simulation data; and the long-term prediction terminal dynamically adjusts the weight of the component, establishes a long-term running simulation model and executes interaction processing. Through the method, the prediction error rate of the key component can be controlled, and the intelligent level of reliability analysis, prediction and management of the nuclear power key equipment is improved.

Description

Technical Field

[0001] This invention relates to the technical field of reliability analysis of nuclear power equipment, and more specifically, to an intelligent reliability analysis system and method for key nuclear power equipment. Background Technology

[0002] The utilization rate of nuclear power equipment condition monitoring data is less than 30%, failure mode analysis relies on expert experience, health assessment lacks dynamic data support, and traditional reliability management systems struggle to integrate equipment disassembly structure, operating parameters, and historical failure data.

[0003] In existing technologies, existing systems cannot achieve dynamic reliability assessment at the equipment component level, fault warnings are delayed, and health assessment models do not consider the disassembly configuration characteristics of equipment, resulting in insufficient prediction accuracy. When specific process queries are required, it is time-consuming and labor-intensive, and the overall level of intelligent management is low. Therefore, this invention needs to design an intelligent analysis system and prediction method for the reliability of key nuclear power equipment to solve the above-mentioned problems. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide an intelligent analysis system and method for the reliability of key nuclear power equipment, addressing the problems existing in the prior art.

[0005] The technical solution adopted by this invention to solve its technical problem is: to construct an intelligent analysis system for the reliability of key nuclear power equipment, including: a knowledge graph construction module, a dual attention mechanism neural network model, a heat map monitoring module, and a long-term prediction terminal;

[0006] The knowledge graph construction module is used to acquire real-time sensor data of key nuclear power equipment, construct an equipment disassembly tree knowledge graph, and store fault mode data.

[0007] The dual attention mechanism neural network model is used to generate a dual attention mechanism model, capture spatial and temporal dimensions, process and extract features from the real-time sensor data to construct a neural network model and a real-time prediction model, and synchronously update and optimize the neural network model and the real-time prediction model.

[0008] The heat map monitoring module is used to dynamically update the target three-dimensional heat map based on the real-time sensor data, monitor the dynamic reliability heat map engine, and perform visual monitoring and processing of fault propagation simulation data.

[0009] The long-term prediction terminal is used to access real-time updated algorithms and dynamically adjust component weights, establish a long-term running simulation model, and perform interactive processing.

[0010] In the intelligent analysis system for the reliability of key nuclear power equipment described in this invention, the knowledge graph construction module includes: a real-time sensor data stream acquisition unit, a structured knowledge graph construction unit, and a fault mode library;

[0011] The real-time sensor data stream acquisition unit is used to acquire the real-time sensor data from multiple sets of sensors in the nuclear power equipment;

[0012] The input end of the structured knowledge graph construction unit is connected to the output end of the real-time sensor data stream acquisition unit, and is used to convert the PBS structure into a heterogeneous graph to obtain the device disassembly tree knowledge graph.

[0013] The fault mode library is used to classify and store the fault mode data.

[0014] In the intelligent analysis system for the reliability of key nuclear power equipment described in this invention, the dual attention mechanism neural network model includes: a dual attention mechanism unit, a neural network model construction unit, and a neural network model synchronization optimization unit;

[0015] The dual attention mechanism unit is used to generate a dual attention mechanism model, which captures space and time separately, and captures both the spatial and temporal dimensions simultaneously.

[0016] The neural network model building unit is used to process and extract features from the real-time sensor data in order to build the neural network model and the real-time prediction model.

[0017] The neural network model synchronization optimization unit is used to perform data synchronization updates and optimizations on the neural network model and the real-time prediction model.

[0018] In the intelligent reliability analysis system for key nuclear power equipment described in this invention, the heat map monitoring module includes: a dynamic reliability heat map engine, a dynamic reliability heat map monitoring unit, and a fault propagation simulation unit;

[0019] The dynamic reliability heatmap engine is used to receive the real-time sensor data in real time and dynamically update the heatmap.

[0020] The dynamic reliability heatmap monitoring unit is used to monitor the real-time operation data of the dynamic reliability heatmap engine and to perform visual monitoring and processing of fault propagation simulation data.

[0021] The fault propagation simulation unit is used to generate a three-dimensional fault propagation engine and to simulate the diffusion path of faults along the PBS topology through visualization.

[0022] In the intelligent analysis system for the reliability of key nuclear power equipment described in this invention, the long-term prediction terminal includes: an adaptive weight allocation algorithm unit, a long-term prediction unit, and a visual interactive terminal;

[0023] The adaptive weight allocation algorithm unit is used to access the real-time updated algorithm and dynamically adjust the component weights;

[0024] The long-term prediction unit is used to establish a simulation model for long-term operation and, based on the working conditions and historical performance of key nuclear power equipment, to simulate and predict the long-term operational reliability of the key nuclear power equipment.

[0025] The visual interactive terminal is used to perform interactive processing.

[0026] In the intelligent analysis system for the reliability of key nuclear power equipment described in this invention, the visual interactive terminal includes any one or more of the following interaction modes: remote interaction, on-site interaction, and mobile interaction.

[0027] In the intelligent analysis system for the reliability of key nuclear power equipment described in this invention, the long-term prediction terminal includes multiple terminals, and each of the multiple long-term prediction terminals is used in conjunction with the knowledge graph construction module, the dual attention mechanism neural network model, and the heat map monitoring module, and the multiple long-term prediction terminals are located in different geographical locations.

[0028] This invention also provides an intelligent reliability analysis method for key nuclear power equipment, applied to the aforementioned intelligent reliability analysis system for key nuclear power equipment, comprising the following steps:

[0029] Step S1. Construct a knowledge graph of equipment disassembly tree and integrate fault mode data and real-time sensor data to fuse the data;

[0030] Step S2. Construct a dual attention mechanism model, capture spatial and temporal dimensions, process and extract features from the real-time sensor data to construct a neural network model and a real-time prediction model, and synchronously update and optimize the neural network model and the real-time prediction model.

[0031] Step S3. Construct a three-dimensional visualization reliability heatmap, dynamically display the component health index based on the PBS structure, and build a fault propagation simulation engine;

[0032] Step S4. Generate a target three-dimensional heat map based on the three-dimensional visualized reliability heat map, perform visual monitoring processing on the fault propagation simulation data based on the fault propagation simulation engine, generate a fault propagation report, and perform visual push.

[0033] In the intelligent reliability analysis method for key nuclear power equipment described in this invention, step S1, which involves fusing the integrated fault mode data and real-time sensor data, includes:

[0034] Align the vibration data, temperature data, and flow data from the real-time sensor data to the PBS node.

[0035] In the intelligent reliability analysis method for key nuclear power equipment described in this invention, in step S2, the spatial dimension includes a spatial attention layer; the temporal dimension includes a temporal attention layer.

[0036] The spatial attention layer includes: device disassembly structural features;

[0037] The time-related layer includes: the time-series correlation of operating parameters.

[0038] The intelligent reliability analysis system and method for key nuclear power equipment of this invention has the following beneficial effects: It includes a knowledge graph construction module, a dual attention mechanism neural network model, a heatmap monitoring module, and a long-term prediction terminal. The knowledge graph construction module constructs a knowledge graph of equipment decomposition trees and stores fault mode data. The dual attention mechanism neural network model generates a dual attention mechanism model, captures spatial and temporal dimensions, constructs a neural network model and a real-time prediction model, and synchronously updates and optimizes the models. The heatmap monitoring module dynamically updates the target three-dimensional heatmap, monitors the dynamic reliability heatmap engine, and performs visual monitoring of fault propagation simulation data. The long-term prediction terminal dynamically adjusts component weights, establishes a long-term running simulation model, and performs interactive processing. This invention can control the prediction error rate of key components and improve the intelligence level of reliability analysis and prediction management for key nuclear power equipment. Attached Figure Description

[0039] The present invention will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings:

[0040] Figure 1 This is a schematic diagram of the intelligent reliability analysis system for key nuclear power equipment provided in an embodiment of the present invention;

[0041] Figure 2 This is a flowchart illustrating the intelligent reliability analysis method for key nuclear power equipment provided in this embodiment of the invention. Detailed Implementation

[0042] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0043] refer to Figure 1 , Figure 1 This paper illustrates a preferred embodiment of the intelligent reliability analysis system for key nuclear power equipment provided by the present invention.

[0044] like Figure 1 As shown, the intelligent reliability analysis system for key nuclear power equipment includes: a knowledge graph construction module 10, a dual attention mechanism neural network model 20, a heat map monitoring module 30, and a long-term prediction terminal 40. The knowledge graph construction module 10 acquires real-time sensor data from key nuclear power equipment, constructs an equipment disassembly tree knowledge graph, and stores fault mode data. The dual attention mechanism neural network model 20 is connected to the knowledge graph construction module 10. It generates a dual attention mechanism model, captures spatial and temporal dimensions, processes and extracts features from real-time sensor data to construct a neural network model and a real-time prediction model, and synchronously updates and optimizes both models. The heat map monitoring module 30 is connected to the knowledge graph construction module 10. It dynamically updates the target 3D heat map based on real-time sensor data, monitors the dynamic reliability heat map engine 301, and performs visual monitoring of fault propagation simulation data. The long-term prediction terminal 40 accesses the real-time updated algorithm, dynamically adjusts component weights, establishes a long-term running simulation model, and performs interactive processing. Optionally, in this embodiment of the invention, the long-term prediction terminal 40 includes multiple terminals, and the multiple long-term prediction terminals 40 are used in conjunction with the knowledge graph construction module 10, the dual attention mechanism neural network model 20, and the heat map monitoring module 30, and the multiple long-term prediction terminals 40 are located in different geographical locations.

[0045] Specifically, the knowledge graph construction module 10 acquires real-time sensor data from multiple sets of sensors in key nuclear power equipment, constructs an equipment breakdown tree knowledge graph, and classifies and stores fault mode data. This fault mode data includes, but is not limited to, fault types, frequencies, impact ranges, and possible solutions discovered during actual operation.

[0046] Optionally, in this embodiment of the invention, the knowledge graph construction module 10 includes: a real-time sensor data stream acquisition unit 101, a structured knowledge graph construction unit 102, and a fault mode library 103. The real-time sensor data stream acquisition unit 101 is used to acquire real-time sensor data from multiple sets of sensors in the nuclear power equipment; these multiple sets of sensors include, but are not limited to, temperature sensors, pressure sensors, radiation detectors, vibration sensors, etc. Correspondingly, the real-time sensor data includes, but is not limited to, temperature data, pressure data, radiation data, vibration data, etc. By monitoring the operating status of key nuclear power equipment through these sensors, and by rapidly and continuously collecting the raw data output by the sensors, real-time basis is provided for subsequent analysis and decision-making. The input end of the structured knowledge graph construction unit 102 is connected to the output end of the real-time sensor data stream acquisition unit 101, and is used to convert the PBS structure into a heterogeneous graph to obtain an equipment disassembly tree knowledge graph; through this structured knowledge graph construction unit 102, the PBS structure is converted into a heterogeneous graph for the first time, realizing the ternary binding of "physical assembly-fault logic-real-time data" to obtain an equipment disassembly tree knowledge graph. Among them, PBS (Plant Breakdown Structure), in the field of nuclear power plant equipment management, is a function-oriented hierarchical tree structure used to systematically decompose and manage the physical entities of a nuclear power plant. Fault Mode Library 103 is used to classify and store fault mode data; through this fault mode library 103, the types, frequencies, impact ranges, and possible solutions of faults discovered in actual operation can be classified and stored, providing basic data for subsequent analysis and prediction.

[0047] In this embodiment of the invention, the dual attention mechanism neural network model 20 is used to generate a dual attention mechanism model, simultaneously capturing the spatial and temporal dimensions, processing and extracting features from real-time sensor data of key nuclear power equipment, constructing a neural network model through preprocessing of sensor data, and synchronously updating and optimizing the two models within the neural network model construction unit 202.

[0048] Optionally, in this embodiment of the invention, the dual attention mechanism neural network model 20 includes: a dual attention mechanism unit 201, a neural network model construction unit 202, and a neural network model synchronization optimization unit 203. The dual attention mechanism unit 201 is used to generate a dual attention mechanism model, capturing spatial and temporal aspects separately, while simultaneously capturing both spatial and temporal dimensions. Specifically, the dual attention mechanism unit 201 generates a dual attention mechanism model that can capture spatial and temporal aspects separately, simultaneously capturing both spatial dimensions (component adjacency relationships) and temporal dimensions (parameter hysteresis effects), thereby improving the accuracy of degradation trend prediction. The spatial attention layer extracts structural features from the disassembled equipment, such as the stress distribution weights of flange bolts; the temporal attention layer captures the temporal correlation of operating parameters, such as temperature-pressure coupling drift. The neural network model construction unit 202 is used to process and extract features from real-time sensor data to construct a neural network model and a real-time prediction model. Specifically, the neural network model building unit 202 processes sensor data (preprocessing, including but not limited to data cleaning, normalization, dimensionality reduction, etc.) and extracts features to extract important features containing the operating status of key nuclear power equipment, thereby providing high-quality input data for the construction of the neural network model. Simultaneously, the neural network model building unit 202 is also used to build a real-time prediction model, which receives and analyzes the operating data of key nuclear power equipment in real time, providing support for real-time decision-making. The neural network model synchronization optimization unit 203 is used to synchronize and optimize the data of the neural network model and the real-time prediction model. Optionally, in this embodiment of the invention, the neural network model synchronization optimization unit 203 can be integrated inside the neural network model building unit 202, enabling synchronized updates and optimization of the two models (the neural network model and the real-time prediction model) within the neural network model building unit 202.

[0049] Specifically, the heat map monitoring module 30 can receive real-time sensor data from key nuclear power equipment and dynamically update the heat map, monitor the real-time operation data of the dynamic reliability heat map engine 301, and perform visual monitoring and processing of fault propagation simulation data.

[0050] Optionally, in this embodiment of the invention, the heatmap monitoring module 30 includes: a dynamic reliability heatmap engine 301, a dynamic reliability heatmap monitoring unit 302, and a fault propagation simulation unit 303. The outputs of both the dynamic reliability heatmap engine 301 and the fault propagation simulation unit 303 are connected to the input of the dynamic reliability heatmap monitoring unit 302. The dynamic reliability heatmap engine 301 receives real-time sensor data, dynamically updates the heatmap, and displays the current operating status and reliability of key nuclear power equipment. Through the dynamic display of the heatmap, abnormal patterns in the operation of key nuclear power equipment can be identified in real time, such as significantly low reliability or sudden changes in certain areas, indicating potential fault signs. The dynamic reliability heatmap monitoring unit 302 monitors the real-time data of the dynamic reliability heatmap engine 301, performs visual monitoring of the fault propagation simulation data of the fault propagation simulation unit 303, and also allows for manual interaction based on the monitoring results. The fault propagation simulation unit 303 generates a three-dimensional fault propagation engine and visualizes the propagation path of the fault along the PBS topology. A three-dimensional fault propagation engine is generated by the fault propagation simulation unit 303, which allows users to click on components to trigger critical path highlighting. The engine visualizes the propagation path of the fault along the PBS topology, thus assisting in maintenance decision-making.

[0051] Specifically, the long-term prediction terminal 40 is used to access real-time updated algorithms, dynamically adjust component weights, and ensure that the prediction error of key components is under control; it also establishes a long-term simulation model to simulate the long-term reliability of the equipment based on its working conditions and historical performance.

[0052] Optionally, in this embodiment of the invention, the long-term prediction terminal 40 includes: an adaptive weight allocation algorithm unit 401, a long-term prediction unit 402, and a visual interactive terminal 403. The adaptive weight allocation algorithm is integrated within the long-term prediction unit 402, and the long-term prediction unit 402 and the visual interactive terminal 403 are bidirectionally connected. The adaptive weight allocation algorithm unit 401 is used to access the real-time updated algorithm and dynamically adjust the component weights, thereby ensuring that the prediction error of key components is under control. The long-term prediction unit 402 is used to establish a long-term operation simulation model and, based on the working conditions and historical performance of key nuclear power equipment, simulate and predict the long-term operational reliability of key nuclear power equipment, thereby reducing reliance on historical data. The long-term prediction unit 402 also combines factors such as the material characteristics, aging process, and external environmental influences of nuclear power equipment to establish an equipment life prediction model, uses the equipment life prediction model to predict the life of key nuclear power equipment, and dynamically adjusts the prediction results through real-time data. The visual interactive terminal 403 is used to perform interactive processing. The visual interactive terminal 403 is used to perform a certain degree of interactive processing during the operation of the intelligent analysis system for the reliability of key nuclear power equipment. It can include any one or more of the following interaction modes: remote interaction, on-site interaction, and mobile interaction. Through interaction, manual intervention can be performed to update data for multiple sets of models required for the operation of the intelligent analysis system for the reliability of key nuclear power equipment.

[0053] In practical applications, users can log in to the long-term prediction terminal 40 to access the intelligent analysis system for the reliability of key nuclear power equipment and view the real-time operating data of the knowledge graph construction module 10, the dual attention mechanism neural network model 20, and the heat map monitoring module 30 one by one. The long-term prediction terminal 40 maintains data communication and sharing with the structured knowledge graph construction module 10, the dual attention mechanism neural network model 20, and the dynamic reliability heat map monitoring module 30 through the Internet of Things.

[0054] refer to Figure 2 The present invention also provides an intelligent analysis method for the reliability of key nuclear power equipment, which can be applied to the aforementioned intelligent analysis system for the reliability of key nuclear power equipment.

[0055] Specifically, such as Figure 2 As shown, in a preferred embodiment, the intelligent reliability analysis method for key nuclear power equipment includes the following steps:

[0056] Step S1. Construct a knowledge graph of equipment disassembly tree and integrate fault mode data and real-time sensor data to fuse the data.

[0057] In this embodiment of the invention, step S1, integrating fault mode data and real-time sensor data to fuse the data includes: aligning the vibration data, temperature data, and flow data in the real-time sensor data to the PBS node.

[0058] Step S2. Construct a dual attention mechanism model, capture spatial and temporal dimensions, process and extract features from real-time sensor data to build a neural network model and a real-time prediction model, and synchronously update and optimize the neural network model and the real-time prediction model.

[0059] In this embodiment of the invention, in step S2, the spatial dimension includes: a spatial attention layer; the temporal dimension includes: a temporal attention layer; the spatial attention layer includes: equipment disassembly structural features; the temporal attention layer includes: the temporal correlation of operating parameters.

[0060] Step S3. Construct a three-dimensional visualization reliability heatmap, dynamically display the component health index based on the PBS structure, and build a fault propagation simulation engine.

[0061] Step S4. Generate a target 3D heat map based on the 3D visualized reliability heat map, perform visual monitoring processing on the fault propagation simulation data based on the fault propagation simulation engine, generate a fault propagation report, and push the data to the system.

[0062] This invention constructs a comprehensive intelligent reliability analysis system for key nuclear power equipment by setting up a knowledge graph construction module 10, a dual attention mechanism neural network model 20, a heat map monitoring module 30, and a long-term prediction terminal 40. During actual operation, it integrates a fault mode library 103 with real-time monitoring data streams by constructing a structured knowledge graph based on equipment disassembly trees (i.e., equipment disassembly tree knowledge graph), fusing the data, developing a dual attention mechanism neural network model 20 to achieve spatiotemporal correlation analysis between equipment configuration features and operating parameters, developing a three-dimensional visualized reliability heat map, dynamically displaying component health indices based on the PBS structure, constructing a fault propagation simulation engine to improve early warning accuracy, and incorporating an innovative equipment disassembly structure weight allocation algorithm to control the prediction error rate of key components. It generates the final three-dimensional heat map and fault propagation report, managing, visualizing, and storing the reliability analysis and prediction data and corresponding analysis results for key nuclear power equipment. This facilitates the realization of reliability analysis and prediction management for key nuclear power equipment through IoT cloud management, improving the intelligence level of reliability analysis and prediction management for key nuclear power equipment.

[0063] The invention will now be illustrated with a specific example.

[0064] In this example, the intelligent reliability analysis system for critical nuclear power equipment is connected to the cooling pumps of the nuclear reactor:

[0065] In step S1: After staff inspect the on-site and remote equipment and confirm their normal startup, the Nuclear Power Key Equipment Reliability Intelligent Analysis System is activated. The structured knowledge graph construction unit 102 constructs a structured knowledge graph based on the equipment disassembly tree, integrating the fault mode library 103 with real-time sensor data streams. Data fusion is performed, aligning vibration data, temperature data, and process data to PBS nodes. Simultaneously, the real-time sensor data stream acquisition unit 101 acquires real-time data from multiple sets of sensors, including temperature sensors, pressure sensors, radiation detectors, and vibration sensors. These sensors monitor the operating status of the cooling pump equipment. Rapid and continuous acquisition of raw sensor output data provides real-time data for subsequent analysis and decision-making. This invention utilizes the structured knowledge graph construction unit 102 to transform the PBS structure into a heterogeneous graph for the first time, achieving a three-element binding of "physical assembly - fault logic - real-time data." Furthermore, the fault mode library 103 categorizes and stores the fault types, frequencies, impact ranges, and possible solutions discovered during actual operation, providing foundational data for subsequent analysis and prediction of the cooling pump.

[0066] In step S2, a dual-attention mechanism neural network model 20 is developed to achieve spatiotemporal correlation analysis between equipment configuration features and operating parameters, generating a dual-attention mechanism model. Specifically, by separately capturing spatial and temporal dimensions, the accuracy of degradation trend prediction is improved. The spatial attention layer extracts disassembled structural features of the equipment, such as the stress distribution weights of flange bolts; the temporal attention layer captures the temporal correlation of operating parameters, such as temperature-pressure coupling drift. The neural network model construction unit 202 processes and extracts features from real-time sensor data to extract important features containing the operating status of the cooling pump equipment, thereby constructing a neural network model by providing high-quality input data. The neural network model construction unit 202 also constructs a real-time prediction model, which receives and analyzes the operating data of the cooling pump equipment in real time, providing support for real-time decision-making. Simultaneously, the neural network model synchronization optimization unit 203 performs data synchronization updates and optimizations on the two models within the neural network model construction unit 202.

[0067] In step S3, a three-dimensional visualized reliability heatmap is constructed. This invention dynamically displays component health indices based on the PBS structure and constructs a fault propagation simulation engine, improving early warning accuracy by 40%. By integrating an innovative equipment disassembly structure weight allocation algorithm, the prediction error rate of key components is controlled within 5%. It can receive sensor data from key nuclear power equipment in real time and dynamically update the three-dimensional visualized reliability heatmap. The three-dimensional visualized reliability heatmap displays the current operating status and reliability of the equipment. Through the dynamic display of the three-dimensional visualized reliability heatmap, abnormal patterns in equipment operation can be identified in real time, such as certain areas showing low reliability or sudden changes, indicating potential fault signs. The dynamic reliability heatmap monitoring unit 302 monitors the real-time operating data of the dynamic reliability heatmap engine 301 and performs visualized monitoring of the fault propagation simulation data. Users can interact based on the monitoring results. The fault propagation simulation unit 303 generates a three-dimensional fault propagation engine, and users can click on components to trigger the highlighting of associated paths. Visual simulation of the fault propagation path along the PBS topology assists in maintenance decision-making.

[0068] In step S4, a visualization push is executed: the final 3D power map (target 3D heat map) and fault propagation report are generated. The adaptive weight allocation algorithm unit 401 connects to a real-time updated algorithm to dynamically adjust component weights, ensuring that prediction errors for critical components are controlled. A long-term prediction unit 402 establishes a long-term simulation model, using this model to simulate and predict the reliability of the cooling pump based on the equipment's operating conditions and historical performance, reducing reliance on historical data. Simultaneously, the long-term prediction unit 402, combined with factors such as the cooling pump's material properties, aging process, and external environmental influences, establishes a cooling pump lifespan prediction model, and dynamically adjusts the prediction results using real-time data. A visualization interactive terminal 403 performs a certain degree of interactive processing during system operation (including remote interaction, on-site interaction, mobile interaction, and other interaction modes), allowing manual data updates for multiple sets of models required for the operation of the intelligent analysis system for the reliability of key nuclear power equipment.

[0069] In practical applications, multiple long-term prediction terminals 40 are used in conjunction with a structured knowledge graph construction module 10, a dual-attention mechanism neural network model 20, and a dynamic reliability heatmap monitoring module 30. These multiple long-term prediction terminals 40 are located in different geographical locations. This invention constructs a comprehensive intelligent reliability analysis system for key nuclear power equipment by setting up the structured knowledge graph construction module 10, the dual-attention mechanism neural network model 20, the dynamic reliability heatmap monitoring module 30, and the long-term prediction terminals 40. During actual operation, a structured knowledge graph based on an equipment disassembly tree is constructed, integrating a fault mode library 103 with real-time monitoring data streams to fuse the data. We developed a dual-attention mechanism neural network model 20 to achieve spatiotemporal correlation analysis between equipment configuration features and operating parameters. By developing a three-dimensional visualization reliability heat map, we dynamically display the component health index based on the PBS structure, construct a fault propagation simulation engine to improve the early warning accuracy, and integrate an innovative equipment disassembly structure weight allocation algorithm to control the prediction error rate of key components. We generate the final three-dimensional heat map and fault propagation report, and manage, visualize, and store the reliability analysis and prediction data and corresponding analysis results of key nuclear power equipment. This helps to realize the reliability analysis and prediction management of key nuclear power equipment through IoT cloud management and control, and improve the intelligence level of reliability analysis and prediction management of key nuclear power equipment.

[0070] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0071] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0072] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0073] The above embodiments are only for illustrating the technical concept and features of the present invention, and are intended to enable those skilled in the art to understand the content of the present invention and implement it accordingly. They do not limit the scope of protection of the present invention. All equivalent changes and modifications made within the scope of the claims of the present invention should fall within the scope of the claims of the present invention.

Claims

1. A nuclear power key equipment reliability intelligent analysis system, characterized in that, include: Knowledge graph construction module, dual attention mechanism neural network model, heatmap monitoring module, and long-term prediction terminal; The knowledge graph construction module is used to acquire real-time sensor data of key nuclear power equipment, construct an equipment disassembly tree knowledge graph, and store fault mode data. The dual attention mechanism neural network model is used to generate a dual attention mechanism model, capture spatial and temporal dimensions, process and extract features from the real-time sensor data to construct a neural network model and a real-time prediction model, and synchronously update and optimize the neural network model and the real-time prediction model. The heat map monitoring module is used to dynamically update the target three-dimensional heat map based on the real-time sensor data, monitor the dynamic reliability heat map engine, and perform visual monitoring and processing of fault propagation simulation data. The long-term prediction terminal is used to access real-time updated algorithms and dynamically adjust component weights, establish a long-term running simulation model, and perform interactive processing.

2. The nuclear power key equipment reliability intelligent analysis system according to claim 1, characterized in that, The knowledge graph construction module includes: a real-time sensor data stream acquisition unit, a structured knowledge graph construction unit, and a fault mode library; The real-time sensor data stream acquisition unit is used to acquire the real-time sensor data from multiple sets of sensors in the nuclear power equipment; The input end of the structured knowledge graph construction unit is connected to the output end of the real-time sensor data stream acquisition unit, and is used to convert the PBS structure into a heterogeneous graph to obtain the device disassembly tree knowledge graph. The fault mode library is used to classify and store the fault mode data. 3.The nuclear power key equipment reliability intelligent analysis system according to claim 1, characterized in that, The dual attention mechanism neural network model includes: a dual attention mechanism unit, a neural network model construction unit, and a neural network model synchronization optimization unit; The dual attention mechanism unit is used to generate a dual attention mechanism model, which captures space and time separately, and captures both the spatial and temporal dimensions simultaneously. The neural network model building unit is used to process and extract features from the real-time sensor data in order to build the neural network model and the real-time prediction model. The neural network model synchronization optimization unit is used to perform data synchronization updates and optimizations on the neural network model and the real-time prediction model.

4. The nuclear power key equipment reliability intelligent analysis system according to claim 1, characterized in that, The heatmap monitoring module includes: a dynamic reliability heatmap engine, a dynamic reliability heatmap monitoring unit, and a fault propagation simulation unit; The dynamic reliability heatmap engine is used to receive the real-time sensor data in real time and dynamically update the heatmap. The dynamic reliability heatmap monitoring unit is used to monitor the real-time operation data of the dynamic reliability heatmap engine and to perform visual monitoring and processing of fault propagation simulation data. The fault propagation simulation unit is used to generate a three-dimensional fault propagation engine and to simulate the diffusion path of faults along the PBS topology through visualization.

5. The nuclear power key equipment reliability intelligent analysis system according to claim 1, characterized in that, The long-term prediction terminal includes: an adaptive weight allocation algorithm unit, a long-term prediction unit, and a visual interactive terminal; The adaptive weight allocation algorithm unit is used to access the real-time updated algorithm and dynamically adjust the component weights; The long-term prediction unit is used to establish a long-term running simulation model, and simulate and predict the long-term running reliability of the nuclear power key equipment based on the working conditions and historical performance of the nuclear power key equipment. The visual interactive terminal is used for executing interactive processing.

6. The nuclear power key equipment reliability intelligent analysis system according to claim 5, characterized in that, The visual interactive terminal comprises any one or more of remote interaction, on-site interaction and mobile interaction modes.

7. The nuclear power key equipment reliability intelligent analysis system according to any one of claims 1-6, characterized in that, The long-term prediction terminal comprises a plurality of long-term prediction terminals, and the plurality of long-term prediction terminals are respectively used in cooperation with the knowledge graph construction module, the double attention mechanism neural network model and the thermal map supervision module, and the plurality of long-term prediction terminals are respectively located at different geographical positions.

8. The nuclear power key equipment reliability intelligent analysis method is applied to the nuclear power key equipment reliability intelligent analysis system in any one of claims 1-6, characterized in that, The method comprises the following steps: Step S1. Constructing a device disassembly tree knowledge graph, and integrating fault mode data and real-time sensor data to fuse the data; Step S2. Constructing a double attention mechanism model, capturing spatial and temporal dimensions, and processing and feature extracting the real-time sensor data to construct a neural network model and a real-time prediction model, and synchronously updating and optimizing the neural network model and the real-time prediction model; Step S3. Constructing a three-dimensional visual reliability thermal map, dynamically displaying a component health index based on a PBS structure, and constructing a fault propagation simulation engine; Step S4. Generating a target three-dimensional thermal map based on the three-dimensional visual reliability thermal map, visually supervising and processing fault propagation simulation data based on the fault propagation simulation engine to generate a fault propagation report and visually pushing. 9.The nuclear power key equipment reliability intelligent analysis method according to claim 8, characterized in that, In the step S1, the integrating fault mode data and real-time sensor data to fuse the data comprises: Aligning vibration data, temperature data and flow data in the real-time sensor data to PBS nodes. 10.The nuclear power key equipment reliability intelligent analysis method according to claim 8, characterized in that, In the step S2, the spatial dimension comprises a spatial attention layer, and the temporal dimension comprises a temporal attention layer; The spatial attention layer comprises device disassembly structure features; The temporal attention layer comprises working condition parameter time sequence correlation.

Images