A nuclear power plant operation health state monitoring system

Abstract

The application discloses a kind of nuclear power equipment operation health state monitoring system, and the application relates to the technical field of nuclear power equipment safety monitoring, including comprehensive monitoring center, the comprehensive monitoring center is connected with the following module, wherein: acoustic emission signal acquisition module is used to obtain waveform data according to the requirement of time difference positioning method and acoustic emission sensor array deployed in service sealed container.The application effectively solves the problem that the defect in complex geometric area such as weld, sealing surface of in-service sealed container is difficult to accurately position by fusing time difference positioning method and genetic algorithm optimized neural network, uses time difference positioning method to provide preliminary position range, reduces neural network search space, significantly improves positioning efficiency and accuracy, genetic algorithm further optimizes neural network structure and parameter, establishes high-precision mapping relationship between signal characteristics and spatial position, so that the system can accurately position early defects such as micro-leakage or micro-crack.

Description

Technical Field

[0001] This invention relates to the field of nuclear power equipment safety monitoring technology, specifically a nuclear power equipment operation health status monitoring system. Background Technology

[0002] Monitoring the operational health status of nuclear power equipment is one of the key technologies to ensure the safe and stable operation of nuclear power. During the operation of a nuclear reactor, the in-service sealing container plays an extremely important safety role. Any small-scale leakage or equipment damage can lead to serious consequences. The sealing container can effectively prevent contamination in the event of a leak. Therefore, by monitoring the operational status of nuclear power equipment (sealing container) in real time, existing risk factors can be identified in a timely manner, and corresponding early warning and remedial measures can be taken to improve the safety and reliability of nuclear energy production.

[0003] In existing technologies, most in-service sealed containers are multi-layered welded structures, and potential leakage sources are easily located in complex geometric areas such as welds and sealing surfaces. Traditional monitoring methods can only achieve overall condition assessment and are difficult to accurately locate defect locations, thus delaying targeted maintenance. Therefore, how to integrate time-of-flight positioning method with neural network optimized by genetic algorithm, using time-of-flight positioning method to provide initial positioning range, narrowing the search space of genetic algorithm, and outputting high-precision defect location based on the mapping relationship between signal features and precise location established by genetic algorithm, so as to improve the accuracy of monitoring the operating health status of in-service sealed containers, is the problem to be solved by this invention. To this end, a nuclear power equipment operating health status monitoring system is proposed. Summary of the Invention

[0004] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a nuclear power equipment operation health status monitoring system, comprising a comprehensive monitoring center, wherein the comprehensive monitoring center is communicatively connected to the following modules, wherein:

[0005] The acoustic emission signal acquisition module is used to acquire waveform data according to the requirements of the time difference positioning method and the acoustic emission sensor array deployed in the in-service sealed container, to ensure that the acoustic emission signal generated by leakage or crack can be effectively captured by at least three or more sensors, and to perform preprocessing operations on the waveform data to extract acoustic emission signal features and integrate them to form an acoustic emission signal feature set, wherein the acoustic emission signal features include signal amplitude, frequency and duration.

[0006] The initial location analysis module analyzes the preliminary location range of the acoustic emission source based on the acoustic emission signal feature data in the acoustic emission signal feature set, and determines the search space of the neural network optimized by the genetic algorithm.

[0007] The fine positioning optimization analysis module is used to fuse initial positioning information with acoustic emission signal characteristics, and establish a mapping relationship between signal characteristics and precise position through a neural network optimized by a genetic algorithm.

[0008] The defect location output module, based on the established mapping relationship between signal characteristics and precise location, outputs high-precision defect location information and displays it on the three-dimensional model of the in-service sealed container through a graphical interface.

[0009] The decision support and early warning module, based on the precise location of potential leak sources in in-service sealed containers, combines a knowledge base and maintenance procedures to automatically recommend targeted maintenance solutions and issue early warning signals to remind relevant personnel to take emergency maintenance measures.

[0010] Preferably, the acoustic emission signal acquisition module specifically includes:

[0011] Based on the internal space structure of the in-service sealed container, multiple acoustic emission sensors are deployed to form an acoustic emission sensor array. According to the time difference positioning method, the original waveform data generated by defect activity is captured synchronously to ensure that leakage or crack signals are effectively captured by at least three acoustic emission sensors.

[0012] The captured raw waveform data is preprocessed, including filtering, denoising, and amplification, to eliminate interference, enhance the signal, and extract acoustic emission signal features, including signal amplitude, frequency, and duration, from the preprocessed waveform data.

[0013] The extracted acoustic emission signal features are integrated to form a standardized acoustic emission signal feature set, which is then stored in the data warehouse of the integrated monitoring center.

[0014] Preferably, the initial positioning analysis module includes a time difference positioning calculation unit and a search space determination unit;

[0015] The time difference positioning calculation unit is used to measure the time difference of the signal arriving at different sensors based on the acoustic emission signal feature data in the acoustic emission signal feature set, and to use the time difference to calculate the preliminary location range of the acoustic emission source, thereby preliminarily determining the area where the potential leakage source is located in the in-service sealed container.

[0016] The search space determination unit is used to determine the search space of the neural network optimized by the genetic algorithm, using the determined preliminary position range as boundary conditions.

[0017] Preferably, the time difference positioning calculation unit specifically includes:

[0018] The absolute timestamps of the same event signal arriving at different acoustic emission sensors are identified and extracted from the acoustic emission signal feature set to ensure the uniformity and synchronization of the time reference;

[0019] Based on the extracted absolute timestamps, the time difference sequence between the arrival of the signal at each acoustic emission sensor is calculated. Combined with the known spatial coordinates of the acoustic emission sensors and the propagation speed of sound waves in the material of the in-service sealed container, the time difference is converted into the spatial distance difference between the sound source and each acoustic emission sensor.

[0020] Using the obtained spatial distance difference information, the preliminary spatial coordinates of the acoustic emission source are calculated by solving the hyperbolic equation system, determining its preliminary location range and error interval, and initially identifying the area where potential leakage sources are located in in-service sealed containers.

[0021] Preferably, the search space determination unit specifically includes:

[0022] The system receives the preliminary location range of the acoustic emission source from the time difference positioning calculation unit, including the spatial coordinates of the acoustic emission source and its error range, as the boundary conditions for spatial constraints.

[0023] Centered on the spatial coordinates of the initial acoustic emission source, and combined with the error interval, a specific geometric region is defined in three-dimensional space as the search space for the genetic algorithm and neural network.

[0024] The mathematical boundary parameters of the constrained search space are passed to the fine localization optimization analysis module, which restricts the initial population generation and search behavior of the genetic algorithm to be carried out within the search space.

[0025] Preferably, the fine positioning optimization analysis module includes a genetic algorithm optimization neural network unit and a model evaluation and adjustment unit;

[0026] The genetic algorithm optimizes the neural network unit, which takes the acoustic emission signal characteristics (signal amplitude, frequency and duration) and a defined search space as input, and uses the genetic algorithm to optimize the neural network. The genetic algorithm continuously adjusts and optimizes the weights and structure of the neural network by simulating natural selection and genetic mechanisms, and establishes a mapping relationship between signal characteristics and precise locations.

[0027] The model evaluation and adjustment unit is used to evaluate the neural network model optimized by the genetic algorithm. By comparing it with known defect location data, the positioning error of the neural network model is calculated. Based on the evaluation results, the parameters of the neural network and the parameters of the genetic algorithm are adjusted and optimized to further improve the accuracy and stability of the neural network model.

[0028] Preferably, the genetic algorithm-optimized neural network unit specifically includes:

[0029] Based on the defined search space, an initial population is generated. Each individual represents a complete neural network model through chromosome encoding, and its gene loci define the topology, connection weights, and bias parameters of the neural network.

[0030] Using the acoustic emission signal feature set as input, the neural network corresponding to each individual is executed, and its prediction error of the defect location is used as the fitness. Based on the fitness value, a selection operator is used to preferentially retain excellent individuals and eliminate individuals with poor performance.

[0031] Crossover and mutation operators are applied to individuals in the population to generate new offspring. By iteratively optimizing the structure and parameters of the neural network, a mapping relationship between signal features and precise locations is established.

[0032] Preferably, the model evaluation and adjustment unit specifically includes:

[0033] Using a reserved dataset of known defects as a validation set, the model is input into the optimized neural network model. By calculating the Euclidean distance between the model's output coordinates and the true coordinates, the average positioning error and standard deviation are quantified to objectively evaluate the model's accuracy and stability.

[0034] Based on the initial evaluation results, the impact of the crossover / mutation rate of the genetic algorithm and the learning rate of the neural network on the model performance is analyzed, and the parameters most sensitive to the localization error are identified, providing direction for targeted adjustments.

[0035] Based on the sensitivity analysis, the configuration of the sensitive parameters is adjusted, and the optimization process is rerun using the adjusted parameters for iterative training and validation until the model's positioning accuracy and stability reach the preset performance threshold.

[0036] Preferably, the defect location output module specifically includes:

[0037] Based on the established mapping relationship between signal features and precise location, the processed waveform data is received, and acoustic emission signal features are extracted for defect location.

[0038] Based on the extracted acoustic emission signal characteristics, the coordinates of potential leakage sources are accurately calculated, and their specific location within the in-service sealed container is determined.

[0039] The coordinates and area information of potential leak sources are displayed intuitively on the 3D model of the in-service sealed container through a graphical interface, and are accurately marked with highlighted and labeled visual elements for operation and maintenance personnel to view.

[0040] Preferably, the decision support early warning module specifically includes:

[0041] It receives precise defect location information of potential leakage sources, retrieves matching maintenance strategy knowledge from the knowledge base, and automatically recommends targeted maintenance solutions based on maintenance strategy knowledge and maintenance procedures, while generating early warning signals.

[0042] Targeted maintenance plans and early warning signals will be communicated to relevant maintenance personnel, reminding them to take emergency maintenance measures based on the maintenance plans and early warnings.

[0043] This invention provides a nuclear power plant operational health status monitoring system. It has the following beneficial effects:

[0044] (I) The nuclear power equipment operation health status monitoring system effectively solves the problem of difficult accurate location of defects in complex geometric areas such as welds and sealing surfaces of in-service sealed containers by integrating time difference positioning method and neural network optimized by genetic algorithm. The time difference positioning method provides an initial location range, which narrows the search space of the neural network and significantly improves the positioning efficiency and accuracy. The genetic algorithm further optimizes the neural network structure and parameters, and establishes a high-precision mapping relationship between signal features and spatial position, enabling the system to accurately locate early defects such as micro-leakage or micro-cracks, and greatly improve the sensitivity and reliability of defect identification.

[0045] (II) This nuclear power equipment operation health status monitoring system introduces an optimization mechanism that combines genetic algorithms and neural networks. It has self-learning and adaptive adjustment capabilities. Through the validation set, error analysis and parameter sensitivity evaluation of the model are carried out to achieve dynamic optimization of key parameters such as crossover rate, mutation rate and learning rate. Through continuous iteration, the model gradually improves its positioning accuracy and stability during continuous operation, forming an intelligent monitoring system with continuous evolution capabilities, which effectively copes with the monitoring challenges brought about by different operating conditions and equipment aging. Attached Figure Description

[0046] Figure 1 This is a schematic diagram of the working process of a nuclear power equipment operation health status monitoring system according to the present invention;

[0047] Figure 2 This is a data flow diagram of a nuclear power equipment operation health status monitoring system according to the present invention. Detailed Implementation

[0048] 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.

[0049] Example 1, please refer to Figure 1 , Figure 2 This invention provides a technical solution: a nuclear power equipment operation health status monitoring system, including a comprehensive monitoring center, which is communicatively connected to the following modules, wherein:

[0050] The acoustic emission signal acquisition module is used to acquire waveform data according to the requirements of the time difference positioning method and the acoustic emission sensor array deployed in the in-service sealed container. This ensures that the acoustic emission signals generated by leakage or cracks can be effectively captured by at least three or more sensors. The module performs preprocessing operations on the waveform data, including filtering, denoising, and amplification, and extracts and integrates acoustic emission signal features to form an acoustic emission signal feature set. The acoustic emission signal features include signal amplitude, frequency, and duration. Based on the internal spatial structure of the in-service sealed container, multiple acoustic emission sensors are deployed to form an acoustic emission sensor array. According to the requirements of the time difference positioning method, the module synchronously captures the original waveform data generated by defect activities, ensuring that leakage or crack signals are effectively captured by at least three acoustic emission sensors. The module performs preprocessing operations on the captured original waveform data, including filtering, denoising, and amplification, to eliminate interference and enhance the signal. The module extracts acoustic emission signal features from the preprocessed waveform data, including signal amplitude, frequency, and duration. The extracted acoustic emission signal features are integrated to form a standardized acoustic emission signal feature set, which is stored in the data warehouse of the integrated monitoring center.

[0051] The specific work of the acoustic emission signal acquisition module is as follows: Based on the internal spatial structure and geometric characteristics of the in-service sealed container, multiple acoustic emission sensors are planned and deployed to form an acoustic emission sensor array covering all monitoring areas of the in-service sealed container. During system operation, according to the requirements of the time difference positioning method, each acoustic emission sensor simultaneously captures the raw waveform data generated by defect activity within the in-service sealed container, ensuring that signals generated by leakage or cracks can be effectively captured by at least three acoustic emission sensors. The raw waveform data captured by the acoustic emission sensors is preprocessed using advanced filtering algorithms to remove high-frequency noise and low-frequency interference from the signal, retaining the effective frequency components related to defect activity, and employing efficient de-interference techniques. Noise reduction technology further eliminates random noise and pulse interference in the signal, improving signal purity. Through amplification, the signal amplitude is enhanced, making the originally weak defect signal clearly visible, eliminating the influence of external interference on the signal, and significantly improving signal quality. From the preprocessed waveform data, signal analysis methods are used to extract acoustic emission signal features, including signal amplitude, frequency, and duration, to reflect the intensity, frequency characteristics, and duration information of defect activity. The extracted acoustic emission signal features are integrated to form a standardized acoustic emission signal feature set, ensuring data consistency and standardization. The integrated acoustic emission signal feature set is then stored in the data warehouse of the integrated monitoring center.

[0052] The initial positioning analysis module analyzes the preliminary location range of the acoustic emission source based on the acoustic emission signal feature data in the acoustic emission signal feature set, and determines the search space of the neural network optimized by the genetic algorithm. The initial positioning analysis module includes a time difference positioning calculation unit and a search space determination unit.

[0053] The time difference positioning calculation unit is used to measure the time difference of the signal arriving at different sensors based on the acoustic emission signal feature data in the acoustic emission signal feature set, and to calculate the preliminary location range of the acoustic emission source using the time difference. This allows for the preliminary determination of the area where potential leakage sources are located in the in-service sealed container. The unit identifies and extracts the absolute timestamps of the same event signal arriving at different acoustic emission sensors from the acoustic emission signal feature set to ensure the uniformity and synchronization of the time reference. Based on the extracted absolute timestamps, the unit calculates the time difference sequence between the signals arriving at each acoustic emission sensor. Combining the known spatial coordinates of the acoustic emission sensors and the propagation speed of sound waves in the material of the in-service sealed container, the unit converts the time difference into the spatial distance difference between the sound source and each acoustic emission sensor. Using the obtained spatial distance difference information, the unit calculates the preliminary spatial coordinates of the acoustic emission source by solving the hyperbolic equation system, determines its preliminary location range and error range, and preliminarily clarifies the area where potential leakage sources are located in the in-service sealed container.

[0054] The specific tasks of the time difference positioning calculation unit are as follows: From the established acoustic emission signal feature set, identify and extract the absolute timestamps corresponding to signals generated on different acoustic emission sensors belonging to the same emission event. Relying on the global synchronization clock system established during the deployment phase, ensure that all acoustic emission sensor data streams have a unified and accurate time reference. By analyzing the waveform characteristics and energy envelope index of the signals, perform event correlation discrimination to confirm acoustic emission sensor signals originating from the same defect activity. After successfully identifying the same event, read the absolute time of arrival of the initial wave point of the signal captured by each acoustic emission sensor, and calculate the time difference between the arrival of the signal at any two sensors, forming a complete time difference sequence. After obtaining the accurate time difference sequence, proceed to the spatial geometric relationship modeling stage, converting the time information into spatial... The distance information is obtained by calling the pre-recorded precise three-dimensional spatial coordinates of the acoustic emission sensor array on the in-service sealed container. Simultaneously, combined with the known propagation speed of sound waves in the in-service sealed container material, and based on wave mechanics principles, the time difference between the signal arrival at the two sensors is equivalent to the spatial path difference between the sound source and the two acoustic emission sensors. By multiplying each calculated time difference by the sound wave propagation speed, it is converted into the spatial distance difference between the sound source and the corresponding acoustic emission sensor pair, forming a hyperbolic equation system with the acoustic emission sensor pair as the focus. Each hyperbola represents the set of all possible locations where the sound source can exist. The established hyperbolic equation system is solved using numerical algorithms to calculate the most probable three-dimensional spatial coordinates of the acoustic emission source, further determining its preliminary location range and error interval, and clarifying the area where the potential leakage source is located.

[0055] The search space determination unit is used to determine the search space of the neural network optimized by the genetic algorithm by using the determined preliminary location range as boundary conditions. This restricts the neural network to establish the mapping relationship between signal features and location within a specific region, avoiding blind searching within the entire in-service sealed container space and narrowing the search range of the genetic algorithm. It receives the preliminary location range of the acoustic emission source output from the time difference positioning calculation unit, including the spatial coordinates of the acoustic emission source and its error range, as the boundary conditions of the spatial constraints. Centered on the preliminary spatial coordinates of the acoustic emission source and combined with the error range, a specific geometric region is defined in three-dimensional space as the search space of the genetic algorithm and the neural network. The mathematical boundary parameters of the constrained search space are passed to the fine positioning optimization analysis module, restricting the initial population generation and search behavior of the genetic algorithm to be carried out within the search space. By limiting the search area of ​​the neural network, it avoids blind searching within the entire in-service sealed container space.

[0056] The specific tasks of the search space determination unit are as follows: It receives the preliminary location range of the acoustic emission source output from the time-difference positioning calculation unit. This range encompasses the spatial coordinates of the acoustic emission source and provides a quantitative description of its approximate location in three-dimensional space. It also includes an error range, reflecting the uncertainty of the positioning results. The preliminary location range serves as the boundary condition for spatial constraints, providing a basis for defining the search space. After obtaining the preliminary location range, it constructs a specific geometric region in three-dimensional space, centered on the preliminary spatial coordinates of the acoustic emission source and incorporating the error range, as the search space. The shape and size of this geometric region are determined based on the size and distribution of the error range, ensuring that it covers the possible real location range of the acoustic emission source, thus defining the originally abstract acoustic emission source. The location range is transformed into an intuitive and operable three-dimensional geometric space, providing a clear physical range for the search of genetic algorithms and neural networks. After defining the search space, the mathematical boundary parameters of the constrained search space are passed to the fine localization optimization analysis module to restrict the search behavior of genetic algorithms and neural networks. For genetic algorithms, the boundary parameters limit the generation range of the initial population, ensuring that the initial population is generated within a reasonable space, thus improving search efficiency. At the same time, they constrain its search behavior to avoid unnecessary searches in invalid regions. For neural networks, the boundary parameters limit their search area, preventing the neural network from blindly searching the entire in-service sealed container space, saving computational resources, accelerating the search speed, and improving the accuracy and reliability of fine localization.

[0057] The fine positioning optimization analysis module is used to fuse initial positioning information with acoustic emission signal characteristics, and establish a mapping relationship between signal characteristics and precise position through a neural network optimized by a genetic algorithm.

[0058] The defect location output module, based on the established mapping relationship between signal characteristics and precise location, outputs high-precision defect location information and intuitively presents the specific coordinates and location information of potential leakage sources in the in-service sealed container. It is displayed on the three-dimensional model of the in-service sealed container through a graphical interface, providing maintenance personnel with clear and accurate defect location information.

[0059] The decision support and early warning module, based on the precise location of potential leak sources in in-service sealed containers, combines a knowledge base and maintenance procedures to automatically recommend targeted maintenance solutions and issue early warning signals to remind relevant personnel to take emergency maintenance measures.

[0060] Example 2, as Figure 1 , Figure 2 As shown, based on Embodiment 1, the present invention provides a technical solution: the fine positioning optimization analysis module includes a genetic algorithm optimization neural network unit and a model evaluation and adjustment unit;

[0061] Genetic algorithms optimize neural network units by taking acoustic emission signal features (signal amplitude, frequency, and duration) and a defined search space as input. The genetic algorithm continuously adjusts and optimizes the weights and structure of the neural network by simulating natural selection and genetic mechanisms, establishing a mapping relationship between signal features and precise locations. Based on the defined search space, an initial population is generated. Each individual represents a complete neural network model through chromosome encoding. Its gene loci define the topology, connection weights, and bias parameters of the neural network. Taking the acoustic emission signal feature set as input, the neural network corresponding to each individual is executed. The prediction error of the defect location is used as the fitness. Based on the fitness value, a selection operator prioritizes retaining superior individuals and eliminates poorly performing individuals. Crossover and mutation operators are applied to individuals in the population to generate new offspring. Through iterative optimization of the neural network's structure and parameters, a mapping relationship between signal features and precise locations is established.

[0062] The specific work of the genetic algorithm in optimizing a neural network unit is as follows: After defining the search space, the genetic algorithm generates an initial population, encodes each individual, and presents a neural network model in the form of a chromosome. It defines the topology of the neural network, including the number of layers and the number of neurons per layer, and specifies the connection weights and bias parameters, transforming the neural network into a heritable chromosome. Using the acoustic emission signal feature set as input, the neural network corresponding to each individual in the population begins operation. The neural network predicts the defect location based on its own structure and parameters. The error between the predicted result and the actual location is used as a fitness index to measure the quality of each individual. Based on the fitness value, selection operators are used to prioritize retaining individuals with high fitness and good predictions. Individuals with small prediction errors are allowed to enter the next generation, while those with poor performance and large prediction errors are eliminated, ensuring that the population evolves in a better direction and gradually approaches a neural network model that can accurately establish the mapping relationship between signal features and precise locations. For the individuals that have been selected and retained in the population, crossover and mutation operators are applied to initiate the process of generating new offspring. The crossover operator exchanges some gene loci on the chromosomes of different individuals to realize the exchange and recombination of information between individuals, creating a neural network structure with new characteristics. The mutation operator randomly changes the values ​​of certain gene loci with probability, introducing new mutation factors and increasing the diversity of the population. Through iterative optimization, the structure and parameters of the neural network are continuously optimized, thereby establishing an accurate and reliable mapping relationship between signal features and precise locations.

[0063] The model evaluation and adjustment unit is used to evaluate the neural network model optimized by the genetic algorithm. By comparing it with known defect location data, the localization error of the neural network model is calculated. Based on the evaluation results, the parameters of the neural network and the genetic algorithm are adjusted and optimized to further improve the accuracy and stability of the neural network model. A reserved known defect dataset is used as a validation set and input into the optimized neural network model. By calculating the Euclidean distance between the model output coordinates and the true coordinates, the average localization error and standard deviation are quantified to objectively evaluate the model's accuracy and stability. Based on the initial evaluation results, the impact of the crossover / mutation rate of the genetic algorithm and the learning rate of the neural network on the model performance is analyzed. The parameters most sensitive to localization error are identified to provide direction for targeted adjustments. Based on the sensitivity analysis, the configuration of sensitive parameters is adjusted. The optimization process is rerun using the adjusted parameters for iterative training and validation until the model's localization accuracy and stability reach the preset performance threshold.

[0064] The specific tasks of the model evaluation and adjustment unit are as follows: Using a reserved known defect dataset as a validation set, the model is input into the optimized neural network model. After the model runs, it outputs the predicted defect location coordinates. By calculating the Euclidean distance between the predicted coordinates and the true coordinates, the model's localization error is quantified. Based on this, the average localization error and standard deviation are further calculated. The average localization error reflects the average deviation between the model's overall predicted position and the true position, while the standard deviation reflects the dispersion of the model's prediction results, i.e., its stability. This objectively evaluates the model's accuracy and stability in the acoustic emission source localization task. Based on the initial model evaluation results, the impact of the crossover rate, mutation rate, and neural network learning rate on model performance is analyzed. The crossover rate and mutation rate determine the degree of information exchange and variation among individuals in the genetic algorithm, directly affecting the population's diversity and evolutionary direction. The learning rate... The step size of parameter updates during neural network training controls the convergence speed and final performance of the model. By changing the values ​​of various parameters, the changes in the model's mean localization error and standard deviation are observed to identify the parameters most sensitive to localization error. Based on the results of sensitivity analysis, the configuration of sensitive parameters is adjusted accordingly. If the crossover rate is found to have a significant impact on localization error, the crossover rate is adjusted appropriately. The genetic algorithm is then re-run using the adjusted parameters to optimize the neural network, and a new round of iterative training and validation is performed. After each iteration, the model's localization accuracy and stability are evaluated again using the validation set. The mean localization error and standard deviation are calculated to determine whether the model's performance has reached the preset performance threshold. If not, the parameters are adjusted and iterations continue until the model's localization accuracy and stability meet the requirements, ensuring that the model can accurately complete the acoustic emission source localization task in practical applications.

[0065] The defect location output module specifically includes: based on the established mapping relationship between signal features and precise location, receiving processed waveform data, extracting acoustic emission signal features for defect location, accurately calculating the coordinates of potential leakage sources based on the extracted acoustic emission signal features, clarifying their specific location in the in-service sealed container, and intuitively displaying the coordinates and area information of potential leakage sources on the 3D model of the in-service sealed container through a graphical interface, and accurately marking them with highlighted and labeled visual elements for maintenance personnel to view;

[0066] The specific tasks of the defect location output module are as follows: Based on the established mapping relationship between signal features and precise locations, the defect location output module receives processed waveform data. The waveform data undergoes preprocessing to remove noise interference. Acoustic emission signal features are extracted from the processed waveform data. After extraction, the coordinates of potential leak sources are initially determined based on the pre-established mapping relationship between signal features and precise locations. After initially determining the coordinates of potential leak sources, the specific location of the potential leak source within the in-service sealed container is clarified. Combining the structural information and internal layout of the in-service sealed container with the initially determined coordinates of the potential leak source, spatial analysis and region division calculations are applied. The method accurately determines the area to which a potential leak source belongs, identifying whether it is located at the top, bottom, side, or a specific functional area of ​​the container. After determining the area, the coordinate information of the potential leak source and the area information are integrated to form a complete and accurate set of defect location information. The coordinates and area information of the potential leak source are then displayed intuitively on the 3D model of the in-service sealed container through a graphical interface. Using visualization technology, a 3D model of the sealed container is constructed and presented in a graphical interface operable by maintenance personnel. On the 3D model, potential leak sources are precisely marked using highlights and labels. Highlighting allows maintenance personnel to quickly focus on the defect location, while labels provide detailed coordinate information and the area to which they belong.

[0067] The decision support and early warning module specifically includes: receiving precise defect location information of potential leakage sources, retrieving matching maintenance strategy knowledge from the knowledge base, and automatically recommending targeted maintenance solutions in combination with maintenance strategy knowledge and maintenance procedures. At the same time, it generates early warning signals and conveys the targeted maintenance solutions and early warning signals to relevant maintenance personnel, reminding them to take emergency maintenance measures based on the maintenance solutions and early warnings.

[0068] The decision support and early warning module's specific tasks are as follows: It receives precise defect location information from potential leak sources and retrieves matching maintenance strategy knowledge from a knowledge base. This knowledge base covers information on various sealed container faults and corresponding maintenance methods. Through precise matching, it quickly identifies effective maintenance strategies for the current potential leak source, providing a solid theoretical basis for subsequent maintenance plan development and ensuring the scientific and targeted nature of maintenance work. After completing the maintenance strategy knowledge matching, it automatically recommends targeted maintenance plans based on the retrieved maintenance strategy knowledge and established maintenance procedures. These procedures are standardized operating processes tested in practice, ensuring the standardization and safety of maintenance work. Simultaneously, it generates early warning signals, providing a direct reflection of the severity of potential problems with the sealed container. These warnings allow maintenance personnel to quickly understand the urgency of the problem, providing clear guidance for subsequent measures and preventing further deterioration. Finally, it promptly transmits the generated targeted maintenance plans and early warning signals to relevant maintenance personnel. While conveying this information, it reminds maintenance personnel to take emergency maintenance measures based on the maintenance plans and warnings, including equipment shutdown for inspection, component replacement, and system debugging, aiming to eliminate potential safety hazards and ensure the normal operation of the equipment.

[0069] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0070] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A nuclear power equipment operational health status monitoring system, comprising a comprehensive monitoring center, characterized in that, The integrated monitoring center has the following communication connection modules, including: The acoustic emission signal acquisition module is used to acquire waveform data according to the requirements of the time difference positioning method and the acoustic emission sensor array deployed in the in-service sealed container, and to preprocess the waveform data to extract acoustic emission signal features and integrate them to form an acoustic emission signal feature set. The initial positioning analysis module analyzes the preliminary location range of the acoustic emission source based on the acoustic emission signal characteristic data and determines the search space of the neural network optimized by the genetic algorithm. The initial positioning analysis module includes a time difference positioning calculation unit and a search space determination unit. The time difference positioning calculation unit is used to measure the time difference of the signal arriving at different sensors based on the acoustic emission signal feature data in the acoustic emission signal feature set, and to use the time difference to calculate the preliminary location range of the acoustic emission source, thereby preliminarily determining the area where the potential leakage source is located in the in-service sealed container. The search space determination unit is used to determine the search space of the neural network optimized by the genetic algorithm, taking the determined preliminary position range as boundary conditions. Specifically, it includes: The system receives the preliminary location range of the acoustic emission source from the time difference positioning calculation unit, including the spatial coordinates of the acoustic emission source and its error range, as the boundary conditions for spatial constraints. Centered on the spatial coordinates of the initial acoustic emission source, and combined with the error interval, a specific geometric region is defined in three-dimensional space as the search space for the genetic algorithm and neural network. The mathematical boundary parameters of the constrained search space are passed to the fine localization optimization analysis module to restrict the initial population generation and search behavior of the genetic algorithm to be carried out within the search space; The fine positioning optimization analysis module is used to fuse initial positioning information with acoustic emission signal characteristics, and establish a mapping relationship between signal characteristics and precise position through a neural network optimized by a genetic algorithm. The defect location output module, based on the established mapping relationship between signal characteristics and precise location, outputs high-precision defect location information and displays it on the three-dimensional model of the in-service sealed container through a graphical interface. The decision support and early warning module, based on the precise defect location of potential leakage sources in in-service sealed containers, combined with a knowledge base and maintenance procedures, recommends maintenance solutions and issues early warning signals.

2. The nuclear power equipment operation health status monitoring system according to claim 1, characterized in that: The acoustic emission signal acquisition module specifically includes: Based on the internal space structure of the in-service sealed container, multiple acoustic emission sensors are deployed to form an acoustic emission sensor array, and the original waveform data generated by defect activity is captured synchronously according to the time difference positioning method. The captured raw waveform data is preprocessed, including filtering, denoising, and amplification, and acoustic emission signal features, including signal amplitude, frequency, and duration, are extracted from the preprocessed waveform data. The extracted acoustic emission signal features are integrated to form a standardized acoustic emission signal feature set, which is then stored in the data warehouse of the integrated monitoring center.

3. The nuclear power equipment operation health status monitoring system according to claim 1, characterized in that: The time difference positioning calculation unit specifically includes: Identify and extract the absolute timestamps of the same event signal arriving at different acoustic emission sensors from the feature set of acoustic emission signals; Based on the extracted absolute timestamps, the time difference sequence between the arrival of the signal at each acoustic emission sensor is calculated. Combined with the known spatial coordinates of the acoustic emission sensors and the propagation speed of sound waves in the material of the in-service sealed container, the time difference is converted into the spatial distance difference between the sound source and each acoustic emission sensor. Using the obtained spatial distance difference information, the preliminary spatial coordinates of the acoustic emission source are calculated by solving the hyperbolic equation system, determining its preliminary location range and error interval, and initially identifying the area where potential leakage sources are located in in-service sealed containers.

4. The nuclear power equipment operation health status monitoring system according to claim 1, characterized in that: The precise positioning optimization analysis module includes a genetic algorithm optimization neural network unit and a model evaluation and adjustment unit; The genetic algorithm optimizes the neural network unit, which is used to optimize the neural network by taking the acoustic emission signal features and the determined search space as input, adjusting and optimizing the weights and structure of the neural network, and establishing a mapping relationship between signal features and precise positions. The model evaluation and adjustment unit is used to evaluate the neural network model optimized by the genetic algorithm, calculate the positioning error of the neural network model by comparing it with known defect location data, and adjust and optimize the parameters of the neural network and the genetic algorithm based on the evaluation results.

5. The nuclear power equipment operation health status monitoring system according to claim 4, characterized in that: The genetic algorithm-optimized neural network unit specifically includes: Based on the defined search space, an initial population is generated. Each individual represents a complete neural network model through chromosome encoding, and its gene loci define the topology, connection weights, and bias parameters of the neural network. Using the acoustic emission signal feature set as input, the neural network corresponding to each individual is executed, and its prediction error of the defect location is used as the fitness. Based on the fitness value, a selection operator is used to preferentially retain excellent individuals and eliminate individuals with poor performance. Crossover and mutation operators are applied to individuals in the population to generate new offspring. By iteratively optimizing the structure and parameters of the neural network, a mapping relationship between signal features and precise locations is established.

6. The nuclear power equipment operation health status monitoring system according to claim 4, characterized in that: The model evaluation and adjustment unit specifically includes: Using a reserved dataset of known defects as a validation set, the model is input into the optimized neural network model. The average positioning error and standard deviation are quantified by calculating the Euclidean distance between the model's output coordinates and the true coordinates. Based on the initial evaluation results, the impact of the crossover / mutation rate of the genetic algorithm and the learning rate of the neural network on the model performance was analyzed, and the parameters most sensitive to localization error were identified. Based on the sensitivity analysis, the configuration of the sensitive parameters is adjusted, and the optimization process is rerun using the adjusted parameters for iterative training and validation until the model's positioning accuracy and stability reach the preset performance threshold.

7. The nuclear power equipment operation health status monitoring system according to claim 1, characterized in that: The defect location output module specifically includes: Based on the established mapping relationship between signal features and precise location, the processed waveform data is received, and acoustic emission signal features are extracted for defect location. Based on the extracted acoustic emission signal characteristics, the coordinates of potential leakage sources are accurately calculated, and their specific location within the in-service sealed container is determined. The coordinates and area information of potential leak sources are displayed intuitively on the 3D model of the in-service sealed container through a graphical interface, and are accurately marked with highlighted and labeled visual elements for operation and maintenance personnel to view.

8. A nuclear power equipment operation health status monitoring system according to claim 7, characterized in that: The decision support early warning module specifically includes: It receives precise defect location information of potential leakage sources, retrieves matching maintenance strategy knowledge from the knowledge base, and automatically recommends targeted maintenance solutions based on maintenance strategy knowledge and maintenance procedures, while generating early warning signals. Targeted maintenance plans and early warning signals will be communicated to relevant maintenance personnel, reminding them to take emergency maintenance measures based on the maintenance plans and early warnings.

Images