Intelligent dynamic adjustment method for alarm priority in nuclear power plant

Abstract

This application proposes an intelligent dynamic adjustment method for alarm priorities in nuclear power plants, comprising: acquiring multiple real-time operating parameters of the nuclear power plant; verifying the data validity of the multiple real-time operating parameters to obtain a set of real-time operating parameters; calculating multiple first fault probabilities corresponding to multiple fault types of the nuclear power plant based on the set of real-time operating parameters; calculating a first alarm priority based on the multiple first fault probabilities; matching the safety rules corresponding to the set of real-time operating parameters in a safety rule base based on the set of real-time operating parameters, and obtaining a second alarm priority based on the level of the safety rules; and processing the first alarm priority and the second alarm priority according to a preset level strategy to obtain the alarm priority of the nuclear power plant.

Description

Technical Field

[0001] This application mainly relates to the field of nuclear power plants, and in particular to a method for intelligent dynamic adjustment of alarm priority in nuclear power plants. Background Technology

[0002] Against the backdrop of global energy structure transformation and the deep integration of artificial intelligence technology, nuclear power plants, as clean, efficient, and stable baseload energy sources, are ushering in unprecedented development opportunities. However, as a typical complex and massive energy system, the safe operation of nuclear power plants highly depends on reliable alarm and fault diagnosis technologies. Currently, nuclear power plant alarm management systems generally adopt a static priority allocation method based on a rule base, that is, various alarm signals are pre-bound to fixed priority rules. When an alarm is triggered, the system classifies it into three levels—high, medium, and low—according to preset rules and presents it to the operator.

[0003] Long-term engineering practice has shown that this technology has significant drawbacks: under abnormal operating conditions, a single initial fault often triggers dozens of related alarms, leading to operators facing extreme processing pressure of 15 to 20 new alarms per minute. Important alarm signals are thus overwhelmed, severely restricting accident response efficiency and decision-making accuracy. Simultaneously, existing systems lack data validity verification of real-time operating parameters and cannot identify data quality issues such as sensor drift and communication interference, further reducing the accuracy and reliability of alarms. Furthermore, current alarm management mechanisms often rely on a single data-driven model for operating condition identification and alarm output, neglecting the integration and constraints of prior knowledge such as nuclear safety rules. This results in the ineffective resolution of the conflict between safety compliance and dynamic adaptability during accident response, leaving the overall reliability of nuclear power plants significantly insufficient. Summary of the Invention

[0004] Based on the above problems, this application proposes a method for determining alarm priority in nuclear power plants, an electronic device, and a computer storage medium, which significantly improves the accuracy and efficiency of power plant fault identification alarm priority output.

[0005] In a first aspect, this application proposes an intelligent dynamic adjustment method for alarm priorities in nuclear power plants, comprising the following steps: acquiring multiple real-time operating parameters of the nuclear power plant; verifying the data validity of the real-time operating parameters to obtain a set of real-time operating parameters; calculating multiple first fault probabilities corresponding to multiple fault types of the nuclear power plant based on the set of real-time operating parameters; determining a first alarm priority based on the multiple first fault probabilities; matching safety rules corresponding to the set of real-time operating parameters in a safety rule base based on the set of real-time operating parameters, and obtaining a second alarm priority based on the level of the safety rules; and processing the first alarm priority and the second alarm priority according to a preset level strategy to obtain the alarm priority of the nuclear power plant.

[0006] In some embodiments, the step of verifying the data validity of the plurality of real-time operating parameters and obtaining a set of real-time operating parameters includes: filtering the plurality of real-time operating parameters to obtain a plurality of reconfiguration real-time operating parameter sets corresponding to a plurality of subsystems in the nuclear power plant; calculating a plurality of reconfiguration error values ​​based on the real-time operating parameters and the plurality of reconfiguration real-time operating parameter sets; obtaining a plurality of credibility scores corresponding to the plurality of subsystems based on the subsystem adjustment coefficients and the plurality of reconfiguration error values; calculating a comprehensive credibility based on the plurality of credibility scores and the plurality of subsystem weights; and filtering a set of real-time operating parameters from the plurality of real-time operating parameters based on the comprehensive credibility and a preset credibility level.

[0007] In some embodiments, the step of calculating the first alarm priority based on the plurality of first fault probabilities includes: obtaining the qualification parameters of the equipment in the nuclear power plant and the real-time operating status parameters of the nuclear power plant; inputting the real-time operating status parameters, the plurality of first fault probabilities, and the qualification parameters into the first safety network model, the second safety network model, and the third safety network model respectively to obtain a first safety parameter, a second safety parameter, and a third safety parameter; taking the maximum value among the first safety parameter, the second safety parameter, and the third safety parameter as a comprehensive safety parameter; and rating the comprehensive safety parameter according to a preset risk level strategy to obtain the first alarm priority.

[0008] In some embodiments, the intelligent dynamic adjustment method for alarm priority in nuclear power plants further includes calculating the confidence level of the first alarm priority, wherein: the first confidence level is calculated based on the plurality of first fault probabilities and the number of the plurality of fault types; a second confidence level is calculated based on a deterministic function, according to a first safety parameter, a second safety parameter, and a third safety parameter; and the confidence level of the first alarm priority is calculated based on the comprehensive confidence level, the first confidence level, the second confidence level, and the corresponding confidence level weights.

[0009] In some embodiments, the step of calculating multiple first fault probabilities corresponding to multiple fault types of the nuclear power plant based on the real-time operating parameter set includes: inputting the real-time operating parameter set into a fault identification network model to determine the multiple first fault probabilities; obtaining the operating parameters of the fault identification network model; obtaining a second fault probability based on the operating parameters of the fault identification network model, wherein the second fault probability is the probability of the fault identification network model failing within a preset time period; calculating a dynamic threshold based on the preset basic threshold, operating adjustment amount, and operating load of the nuclear power plant; determining whether the second fault probability is greater than the dynamic threshold, and if the determination is no, then accepting the multiple first fault probabilities, and if the determination is yes, then not accepting the multiple first fault probabilities.

[0010] In some embodiments, the intelligent dynamic adjustment method for alarm priority in nuclear power plants further includes: outputting an explanation statement for the alarm priority of the nuclear power plant, wherein, based on the real-time operating parameter set and the nuclear power plant baseline parameters, the contribution of each parameter type in the real-time operating parameter set is calculated; the parameter types are sorted according to their contribution, and the top N parameter types are selected in descending order, wherein N is greater than or equal to 5 and less than the total number of parameter types; based on the fault type and the alarm priority of the nuclear power plant, a retrieval algorithm is used to retrieve corresponding fault handling measures in a knowledge base; and based on the fault type, the alarm priority, the contribution of the N parameter types, and the fault handling measures, an explanation statement corresponding to the alarm priority of the nuclear power plant is generated.

[0011] In some embodiments, the preset level strategy includes: when the safety rule is at the first level, using the second alarm priority as the alarm priority of the nuclear power plant; when the safety rule is at the second level, calculating the alarm priority of the nuclear power plant based on the first alarm priority and the second alarm priority, as well as the corresponding first preset weight and second preset weight; when the safety rule is at the third level, using the first alarm priority as the alarm priority of the nuclear power plant.

[0012] In some embodiments, the preset level strategy includes: determining the confidence interval in which the confidence level of the first alarm priority lies; and determining the alarm priority of the nuclear power plant based on the confidence interval in which the confidence level of the first alarm priority lies and the level of the safety rule, wherein each confidence interval corresponds to multiple levels of the same safety rule.

[0013] Secondly, this application also proposes an electronic device comprising: a memory for storing instructions executable by a processor; and a processor for executing the instructions to implement the method as described in the first aspect.

[0014] Thirdly, this application also proposes a computer storage medium storing computer program code that, when executed by a processor, implements the method described in the first aspect.

[0015] Compared with the prior art, the beneficial effects of this application are as follows: (1) By verifying the validity of real-time operating parameters, unreliable data interference is effectively eliminated, significantly improving the input quality of fault probability calculation, thereby enhancing the accuracy and robustness of alarm priority output.

[0016] (2) By constructing a deep integration mechanism of multiple fault type probabilities and safety rules, the adaptability conflict between static rules and actual working conditions is effectively resolved, and the organic unity of nuclear safety regulatory compliance and dynamic fault response is achieved, which significantly improves the reliability and engineering adaptability of alarm priority determination.

[0017] (3) Based on the comprehensive output of multiple fault types probability, a unique alarm priority is output, which reduces the number of alarms from the source, effectively alleviates the information processing burden of operators, and significantly improves human factors work efficiency and accident response capabilities. Attached Figure Description

[0018] The accompanying drawings are included to provide a further understanding of this application; they are incorporated into and constitute a part of this application. The drawings illustrate embodiments of this application and, together with this specification, serve to explain the principles of this application. In the drawings: Figure 1 This is a schematic diagram of the intelligent dynamic adjustment method for alarm priority in nuclear power plants provided in the embodiments of this application; Figure 2 This is a flowchart illustrating the method for verifying data validity in the intelligent dynamic adjustment method for alarm priority in nuclear power plants provided in this application embodiment; Figure 3 This is a flowchart illustrating the method for obtaining the first alarm priority in the intelligent dynamic adjustment method for alarm priority in nuclear power plants provided in this application embodiment; Figure 4This is a flowchart illustrating the detection network model method in the intelligent dynamic adjustment method for alarm priority in nuclear power plants provided in this application embodiment; Figure 5 This is a flowchart illustrating the method for calculating the confidence level of the first alarm priority in the intelligent dynamic adjustment method for alarm priorities in nuclear power plants provided in this application embodiment; Figure 6 This is a flowchart illustrating the method for generating interpretation statements in the intelligent dynamic adjustment method for alarm priority in nuclear power plants provided in this application embodiment; Figure 7 This is a flowchart illustrating the intelligent dynamic adjustment method for alarm priority in nuclear power plants provided in the embodiments of this application; Figure 8 This is a block diagram of the electronic device provided in the embodiments of this application. Detailed Implementation

[0019] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are merely some examples or embodiments of this application. For those skilled in the art, these drawings can be applied to other similar scenarios without creative effort. Unless obvious from the context or otherwise specified, the same reference numerals in the drawings represent the same structures or operations.

[0020] As indicated in this application, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" do not specifically refer to the singular and may also include the plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of explicitly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0021] Unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps described in these embodiments do not limit the scope of this application. It should also be understood that, for ease of description, the dimensions of the various parts shown in the drawings are not drawn to actual scale. Techniques, methods, and devices known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and devices should be considered part of the specification. In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values. It should be noted that similar reference numerals and letters in the following drawings denote similar items; therefore, once an item is defined in one drawing, it need not be further discussed in subsequent drawings.

[0022] In the description of this application, it should be understood that the orientation or positional relationship indicated by directional terms such as "front, back, up, down, left, right", "horizontal, vertical, horizontal" and "top, bottom" is usually based on the orientation or positional relationship shown in the accompanying drawings, and is only for the convenience of describing this application and simplifying the description. Unless otherwise stated, these directional terms do not indicate or imply that the device or element referred to must have a specific orientation or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation on the scope of protection of this application; the directional terms "inner" and "outer" refer to the inner and outer contours relative to the outline of each component itself.

[0023] For ease of description, spatial relative terms such as "above," "on top of," "on the upper surface of," "above," etc., are used herein to describe the spatial positional relationship of a device or feature as shown in the figures to other devices or features. It should be understood that spatial relative terms are intended to encompass different orientations in use or operation beyond the orientation of the device as described in the figures. For example, if the device in the figures were inverted, a device described as "above" or "on top of" other devices or structures would subsequently be positioned as "below" or "under" other devices or structures. Thus, the exemplary term "above" can include both "above" and "below." The device may also be positioned in other different ways (rotated 90 degrees or in other orientations), and the spatial relative descriptions used herein will be interpreted accordingly.

[0024] Furthermore, it should be noted that the use of terms such as "first" and "second" to define components is merely for the purpose of distinguishing the corresponding components. Unless otherwise stated, these terms have no special meaning and therefore should not be construed as limiting the scope of protection of this application. In addition, although the terminology used in this application is selected from commonly known and used terms, some terms mentioned in this application's specification may have been chosen by the applicant according to his or her judgment, and their detailed meanings are explained in the relevant sections of this description. Moreover, this application should be understood not only through the actual terms used, but also through the meaning implied by each term.

[0025] Flowcharts are used in this application to illustrate the operations performed according to embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, various steps can be processed in reverse order or simultaneously. Furthermore, other operations may be added to these processes, or one or more operations may be removed from these processes.

[0026] The following specific embodiments illustrate the intelligent dynamic adjustment method for alarm priority in nuclear power plants according to this application.

[0027] refer to Figure 1 The intelligent dynamic adjustment method for alarm priorities in nuclear power plants includes: Figure 1 Steps S101 to S105 are explained in detail below.

[0028] In step S101, multiple real-time operating parameters of the nuclear power plant are acquired.

[0029] Nuclear power plants deploy various sensors to acquire real-time operating parameters. These sensors can be deployed in both the nuclear island and conventional island to detect operating parameters such as temperature, pressure, and flow rate. They can connect to the nuclear power plant via 4-20mA analog signals or fieldbus. Nuclear power plants can also deploy nuclear-specific monitoring equipment, including nuclear-grade safety monitoring devices such as neutron flux detectors and gamma dose rate monitors. These can be connected via hardwiring, with wiring standards meeting the IEEE 7-4.3.2 nuclear power plant bus standard. Furthermore, nuclear power plants can deploy equipment status sensors, including vibration, displacement, and current sensors, to assess the condition of rotating and electrical equipment.

[0030] Through collaborative monitoring and differentiated access using multiple types of sensors, the operating status of nuclear power plants can be comprehensively perceived, improving safety and predictive maintenance capabilities. Sensors in nuclear power plants can collect real-time operating parameters in real time or at preset frequencies (e.g., at 1-second sampling intervals).

[0031] In some embodiments, after acquiring multiple real-time operating parameters of a nuclear power plant, the method further includes preprocessing the acquired real-time operating parameters.

[0032] First, multiple real-time operating parameters are cross-validated and their rationality checked. Cross-validation utilizes the redundant configuration of key parameters in nuclear power plants and the physical connections between upstream and downstream processes to perform multi-source consistency verification of sensor data. Specifically, for multiple redundant sensors at the same measuring point, the system compares reading deviations in real time. When a sensor deviates from the group mean by more than a preset threshold, it is considered abnormal. For measuring points without direct redundancy, verification is performed through upstream and downstream related parameters. For example, under steady-state conditions, feedwater flow and steam flow should remain balanced; if the deviation exceeds the allowable range, the data is deemed unreliable. Through this cross-validation fusion of multi-source information, the system can identify faulty sensors and select the most reliable data source for subsequent analysis.

[0033] The rationality check involves verifying the inherent logic of sensor readings item by item based on physical laws, historical operating data, and equipment characteristics. Specifically, this includes physical range checks (ensuring values ​​do not exceed possible extreme values), rate of change checks (preventing signal jumps), physical constraint checks between parameters (e.g., using the gas law to verify the relationship between pressure and temperature), and equipment state consistency checks (e.g., the outlet pressure should not be higher than the inlet pressure when the pump is stopped). Data that passes the checks is marked as "reliable" and can be used as input for subsequent models. Slight deviations are weighted less, while significantly abnormal data is replaced with historical values ​​or model predictions, triggering alarms to ensure sufficient reliability of the data input to subsequent models.

[0034] Then, the edge computing gateway of the nuclear power plant performs data denoising and outlier removal operations. The edge computing gateway has a built-in digital filtering algorithm to smooth the raw real-time operating parameters collected by the sensors, filtering out noise components introduced by electromagnetic interference, environmental noise or sensor jitter. At the same time, based on the physical threshold range, the rate of change limit criterion and the statistical anomaly discrimination rules, the outlier removal operation verifies the real-time collected data point by point, and removes data points that exceed the reasonable range or have abrupt anomalies.

[0035] Finally, the cleaned data is segmented using a sliding window mechanism, and statistical features are calculated within each window. Simultaneously, these statistical features are transformed to the frequency domain using methods such as Fast Fourier Transform, thereby comprehensively characterizing the variation patterns of real-time operating parameters of the nuclear power plant in both time and frequency dimensions, providing highly discriminative feature inputs for subsequent fault diagnosis. The preprocessed real-time operating parameters are represented in matrix form, where the matrix represents different types of real-time operating parameters. The preprocessing delay is controlled within 50ms to ensure real-time performance. This preprocessing improves data compatibility and model accuracy, and reduces the complexity of subsequent data processing.

[0036] After the above data preprocessing, multiple real-time operating parameters form a parameter matrix. Each row of the matrix represents a type of real-time operating parameter, such as temperature or pressure. Each column corresponds to a sampling time. The sampling interval can be 1 second, and the number of columns can be 10, 20, 30, or 60.

[0037] In some embodiments, real-time operating parameters acquired at the nuclear power plant are stored in a real-time database. The real-time database can be an OSIsoft PI System, which stores real-time operating parameters with timestamps.

[0038] In step S102, the validity of multiple real-time operating parameters is verified to obtain a set of real-time operating parameters.

[0039] Before utilizing multiple real-time operating parameters, their validity must be verified. Multiple real-time operating parameters that pass validity verification are constructed into a real-time operating parameter set. The validity of these parameters is verified using a parameter matrix. Given a sampling interval of 1 second and a sampling duration of 60 seconds, all sampled data within 60 seconds are organized into a parameter matrix. The data validity verification process involves filtering the parameter matrices formed at different sampling times.

[0040] In some embodiments, reference Figure 2 Based on the verification results, the set of real-time operating parameters is selected, specifically including the following steps S201 to S205.

[0041] In step S201, multiple real-time operating parameters are filtered to obtain multiple sets of reconfigurable real-time operating parameters corresponding to multiple subsystems in the nuclear power plant.

[0042] Nuclear power plants have multiple operating subsystems, such as the core monitoring subsystem, primary loop subsystem, secondary loop subsystem, and dedicated safety facility subsystem. For each operating subsystem, an anomaly detection model is trained, which can be a variational autoencoder. The preprocessed real-time operating parameters are input into the variational autoencoders corresponding to multiple subsystems to obtain multiple reconstructed sets of real-time operating parameters.

[0043] In step S202, multiple reconstruction error values ​​are calculated based on multiple real-time operating parameters and multiple sets of real-time reconstruction operating parameters.

[0044] Multiple reconstruction error values ​​between multiple real-time operating parameters and multiple sets of reconstructed real-time operating parameters are calculated using the reconstruction error formula. The reconstruction error values ​​are calculated according to the following formula (1): (1) E represents the reconstruction error value. The preprocessed real-time operating parameters include different types of real-time operating parameters. This represents a class of real-time operating parameters at the current sampling time. This is the general category for real-time running parameters. This is a type of reconstructed real-time operating parameter obtained by a variational autoencoder from a type of real-time operating parameter.

[0045] In step S203, multiple confidence scores corresponding to multiple subsystems are obtained based on the subsystem adjustment coefficients and multiple reconstruction error values.

[0046] The credibility scores of multiple subsystems are calculated according to formula (2), as follows: (2) Score the credibility of the subsystem. This is the adjustment factor for the subsystem. Different values ​​are set according to the importance of the subsystem. The value range is usually (0, 5]. Important subsystems, such as reactor protection systems, use larger k values, such as 3-5, while minor subsystems use smaller k values, such as 1-2.

[0047] In step S204, the overall credibility is calculated based on multiple credibility scores and multiple subsystem weights.

[0048] The overall credibility is calculated according to formula (3), as follows: (3) It is the overall credibility, with a value range of [0,1]. It represents the overall credibility of the real-time operating parameters of the nuclear power plant at the current moment. The closer its value is to 1, the higher the credibility of the real-time operating parameters. N is the total number of subsystems. The subsystem weight ranges from (0, 10), reflecting its relative importance in the credibility assessment. Subsystems related to nuclear safety and those with good stability have higher weights.

[0049] In step S205, a set of real-time operating parameters is selected from multiple real-time operating parameters based on the overall credibility and the preset credibility level.

[0050] The preset confidence level can be set as follows: if the overall confidence level is greater than 0.7, then the current multiple real-time running parameters are a set of trusted parameters; if the overall confidence level is less than or equal to 0.3, then the current multiple real-time running parameters are a set of untrusted parameters; if the overall confidence level is less than or equal to 0.7 but greater than 0.3, then the current multiple real-time running parameters are marked as a set of warning real-time running parameters. The multiple real-time running parameters marked as trusted parameters are the filtered set of real-time running parameters.

[0051] If the real-time operating parameter set is identified as an untrusted parameter set, the untrusted parameter set is retrieved from the quality diagnostic knowledge base to obtain the reasons for its untrustworthiness. The quality diagnostic knowledge base stores the mapping relationship between various anomaly patterns and possible causes. The input real-time operating parameters undergo validity verification, significantly improving the accuracy and robustness of real-time operating parameter quality assessment and providing a reliable input basis for subsequent calculations of failure probabilities.

[0052] Back Figure 1 In step S103, based on the verified set of real-time operating parameters, multiple first fault probabilities corresponding to multiple fault types in the nuclear power plant are calculated.

[0053] Nuclear power plants may experience various failures during actual operation, such as loss-of-coolant accidents, steam generator heat transfer tube ruptures, feedwater system failures, main pump failures, control rod drive mechanism failures, and core cooling loss. The total number of failure types is predetermined, for example, set to 15. The multiple real-time operating parameters input to this model are validated real-time operating parameter sets, which are parameter matrices composed of multiple real-time operating parameters collected over 60 seconds.

[0054] A validated set of real-time operating parameters is input into a fault identification network model to calculate multiple first fault probabilities corresponding to various fault types. The fault identification network model can employ a Transformer model based on a multi-head attention mechanism. First, the real-time operating parameter set is processed using Transformer encoding; then, it is calculated using the Softmax function of the model's output layer; finally, multiple first fault probabilities corresponding one-to-one with various fault types are output.

[0055] Multiple first-fault probabilities constitute a probability vector. The dimension of this vector is the same as the total number of fault types, for example, 15, and the sum of all first-fault probabilities in this vector is 1. In this way, the attention mechanism is used to capture the complex spatiotemporal correlation between multiple parameters, so as to achieve accurate calculation of fault probabilities.

[0056] In some embodiments, a temporal convolutional network or a long short-term memory network (LSTM) can be used to replace the Transformer model with a multi-head attention mechanism as the fault identification network model to calculate the first fault probability.

[0057] In step S104, a first alarm priority is calculated based on multiple first fault probabilities to determine the multiple first fault probabilities.

[0058] In some embodiments, reference Figure 3 A method for determining the first alarm priority of multiple first fault probabilities based on multiple first fault probabilities includes steps S301 to S304: In step S301, the qualified parameters of the equipment in the nuclear power plant and the real-time operating status parameters of the nuclear power plant are obtained.

[0059] In nuclear power plants, equipment qualification parameters refer to quantitative assessments of the remaining life or health status of critical equipment. These parameters typically come from equipment condition monitoring systems. For example, the health of a main pump is assessed based on vibration, temperature, and operating time, with a score of 0.85 (85% healthy). The health of a feedwater pump is assessed based on bearing wear and current fluctuations, with a score of 0.92. These qualification parameters help assess the reliability of the equipment itself. Real-time operating status parameters of a nuclear power plant refer to the current operating status of the plant. This is a higher-level set of features than the real-time operating parameters obtained from sensors, including operating modes (such as startup, power operation, shutdown, etc.), the plant's power level, the operating status of critical equipment (such as main pump start-up / shutdown, valve opening status), time information (such as day / night, season), and operator workload indicators.

[0060] In step S302, the real-time operating status parameter S, multiple first fault probabilities P, and qualification parameter H are respectively input into the first safety network model. Second security network model and the third security network model Obtain the first security parameter Second safety parameter and third safety parameters .

[0061] Among them, the first security network model Second security network model used to assess reactive control threats. Third Security Network Model for assessing core cooling threats Assess the threat of radioactive containment. The three security network models are represented by the following formulas (4) to (6): (4) (5) (6) in, , and Three deep Q-networks were used. Deep Q-networks are a type of reinforcement learning network used to learn the long-term cumulative reward of taking an action in a given state. State-action-reward samples were constructed using historical operational data and incident simulation data. The reward function was designed such that a positive reward was given if the threat level output by the network matched the actual subsequent consequences, and a negative reward was given if the threat was underestimated. Through repeated training, the network learned to accurately assess threats. The three deep Q-networks used different state-action-reward samples and were trained differently for the threat scenarios they were assessing.

[0062] In step S303, the first safety parameter is taken. Second safety parameter and third safety parameters The maximum value in is used as the comprehensive safety parameter. .

[0063] A comprehensive assessment of the three security threats mentioned above was conducted to obtain comprehensive security parameters. The comprehensive safety parameters are calculated using the following formula (7). : (7) In step S304, the comprehensive safety parameters are adjusted according to the preset risk level strategy. The alarm is rated and assigned the highest priority.

[0064] Specifically, the preset risk level strategy is as follows: when the comprehensive safety parameters... When the value is less than 0.3, the nuclear power plant is considered to be operating at a low risk. This is based on comprehensive safety parameters. When the value is greater than or equal to 0.3 and less than 0.7, the nuclear power plant's operating status is considered medium risk. (This is based on the comprehensive safety parameters.) When the risk level is greater than or equal to 0.7, the nuclear power plant is considered to be in a high-risk operating state. Based on these risk levels, a suggested priority level for the first alarm is obtained: low risk is mapped to a low priority level of the first alarm priority, medium risk to a medium priority level, and high risk to a high priority level. Mapping the failure probability results to the assessment of the three major functional threats to nuclear safety overcomes the limitations of relying solely on failure probability. While accurately obtaining failure identification results, it also considers nuclear safety assessment, outputting a more reliable alarm priority. This compensates for the shortcomings of purely probabilistic models in insufficiently considering the severity of consequences, significantly improving the accuracy, robustness, and safety of nuclear power plant accident response, and reducing the risk of false alarms and missed alarms.

[0065] Continue to refer to Figure 1 In step S105, based on the real-time operating parameter set, the safety rules corresponding to the real-time operating parameter set are matched in the safety rule base, and the second alarm priority is obtained according to the level of the safety rule.

[0066] Specifically, nuclear power plants pre-store a safety rule base, which can be a knowledge base containing 500 safety rules. Within this base, safety rules are divided into three levels. Level 1 rules are absolutely mandatory, including reactor protection rules, dedicated safety facility rules, and nuclear radiation safety rules. For example, when an emergency reactor shutdown signal is detected, all shutdown-related alarms are forcibly set to the highest priority, based on nuclear safety regulations requiring shutdown-related operations to be handled with the highest priority. Level 2 rules are strongly constrained, including rules for critical equipment and process system limitations. For example, when a single coolant pump is detected operating and its vibration value exceeds a threshold, the priority of the relevant first alarm is increased by at least one level, based on equipment protection requirements to prevent system failure due to the failure of the only operating equipment. Level 3 rules are recommended rules, including operational optimization suggestions and maintenance guidance suggestions. For example, when a non-critical system alarm is detected during low-load operation at night, the alarm priority can be appropriately reduced to avoid disturbing operator rest, based on human factors engineering optimization suggestions.

[0067] Based on the real-time operating parameter set, a matching algorithm is used to match the corresponding security rules in the security rule base. The Rete algorithm can be used, which avoids redundant matching calculations and achieves efficient pattern matching by constructing a rule network. The matching result may simultaneously match multiple rules at different levels. When level conflicts occur, the first-level rule (i.e., the highest priority rule) is prioritized as the second alarm priority.

[0068] In step S106, the first alarm priority and the second alarm priority are processed according to the preset level strategy to obtain the alarm priority of the nuclear power plant.

[0069] In some embodiments, the preset level strategy includes: when the safety rule is level 1, using the second alarm priority as the alarm priority of the nuclear power plant; when the safety rule is level 2, calculating the alarm priority of the nuclear power plant based on the first alarm priority and the second alarm priority, and their corresponding first preset weights and second preset weights; when the safety rule is level 3, using the first alarm priority as the alarm priority of the nuclear power plant. Specifically, when the safety rule is level 2, the method for calculating the alarm priority of the nuclear power plant is as follows: the high priority in the first alarm priority and the first level rule in the second alarm priority can be set to 3, the medium priority and the second level rule can be set to 2, and the low priority and the third level rule can be set to 1. The first preset weight and the second preset weight range from (0, 1), and their values ​​can be adjusted according to actual conditions. For example, when the first alarm priority is medium priority, its value can be set to 2; when the second alarm priority is a level 2 rule, its value can be set to 2; if the level 2 rule suggests raising the first alarm priority by one level, then the value of the first alarm priority is 3; and since both the first preset weight and the second preset weight are 0.5, the calculated value is 2.5, which, after rounding, becomes 3, thus the obtained alarm priority of the nuclear power plant is high priority. Thus, the above method, through the fusion mechanism of the first alarm priority output by fault probability and the second alarm priority output by the safety rule base, resolves the conflict between safety rules and adaptability to actual operating conditions, achieving unity between safety regulation compliance and actual fault response, and significantly improving the reliability and accuracy of alarm priority determination. Furthermore, outputting an alarm priority using the fault probabilities of multiple fault types reduces the number of alarms, decreases operator workload, and improves work efficiency. Simultaneously, by constructing a three-layer operating mode including data validity verification, first alarm priority output, and second alarm priority output, the depth, reliability, and robustness of the nuclear power plant alarm priority method are further enhanced.

[0070] In some embodiments, while calculating multiple first alarm priorities based on first fault probabilities, the model used to obtain the first alarm priorities is also monitored, such as a fault identification network model and an anomaly detection model. The monitoring can be performed in real time. If the model passes the monitoring, its output alarm priorities for the nuclear power plant are used; if the monitoring fails, the alarm priorities recommended in a pre-stored expert rule base within the nuclear power plant are used.

[0071] refer to Figure 4 In some embodiments, the model detection process includes the following steps S401 to S405.

[0072] In step S401, the real-time running parameter set is input into the fault identification network model to calculate and obtain multiple first fault probabilities.

[0073] The fault identification network model is a Transformer model that employs a multi-head attention mechanism.

[0074] In step S402, the operating parameters of the fault identification network model are obtained.

[0075] This includes obtaining the attention weight entropy value of the fault identification network model. During computation, the information entropy of the attention weights measures the degree of concentration of attention distribution. A higher entropy value indicates more dispersed attention and unclear focus on key features; a lower entropy value indicates that attention is concentrated on a few features, leading to clearer judgments. Furthermore, the input parameters also include the operating status parameters of the nuclear power plant.

[0076] In step S403, the second fault probability is obtained based on the operating parameters of the fault identification network model.

[0077] The second failure probability is the probability that the fault identification network model will fail within a preset time. The operating parameters of the fault identification network model and the operating status parameters of the nuclear power plant are input into the detection network model. The detection network model can be a gradient boosting decision tree, composed of multiple decision trees. Preferably, the number of decision trees can be 100. Each tree learns the prediction residual of the previous tree, and the overall prediction accuracy is gradually improved through iterative optimization. Each tree has a depth of 6 layers and is trained using 60% of randomly selected features. The output of the detection network model is the second failure probability. This indicates the probability that the detected fault identification network model will fail within the next 30 seconds. This detection network model will be burned into the firmware of the safety PLC.

[0078] In step S404, the dynamic threshold is calculated based on the nuclear power plant's preset basic threshold, operational adjustment amount, and operating load. .

[0079] The current dynamic threshold is a real-time threshold used to determine whether a second fault probability triggers a switchover. Adjustments to the monitoring network model are necessary under different operating conditions at the nuclear power plant. The reliability of the model varies depending on the conditions. During startup, parameters fluctuate drastically, and model training data may be insufficient; therefore, the threshold should be lowered to improve monitoring sensitivity. After a fault or accident, the nuclear power plant system is in an emergency state, prioritizing continuity; the threshold can be appropriately increased to avoid unnecessary switching. Under high load operation, system pressure is high, but parameters are generally more stable; the threshold can be slightly relaxed. Under low load operation, some equipment shuts down, and parameters may fall into areas where model training is insufficient; therefore, the threshold should be lowered. Dynamic threshold. The value range is usually [0.5, 1.0], and it can be calculated using the following formula (8): (8) in, The base threshold is the default threshold for a nuclear power plant system under standard operating conditions, typically set to 0.8. This value is determined based on historical data analysis and the reliability of the nuclear power plant system. It ensures that the nuclear power plant is neither overly sensitive, leading to frequent false switching, nor overly sluggish, leading to missed switching, during normal operation. During the startup phase of the nuclear power plant, the value decreases by 0.2; after a fault occurs, the value increases by 0.1; under high load conditions, the value increases by 0.05; and under low load conditions, the value decreases by 0.05. This is an adjustment amount for the operating mode, and its value range is typically [-0.2, +0.1]. The adjustment method is as follows: during the startup phase, its value decreases by 0.2; during the shutdown and refueling phase, its value decreases by 0.1; and after a failure or accident, its value increases by 0.1. This is the load adjustment amount for nuclear power plant systems, and its value range is typically [-0.05, +0.05]. Under high load (>90% of rated power), its value increases by 0.05; under medium load (30%-90%), its value remains unchanged; and under low load (<30%), its value decreases by 0.05.

[0080] In some embodiments, dynamic threshold It can also include the time factor. Weather factors Further expand the dynamic threshold. The included operating conditions.

[0081] In step S405, the second fault probability is determined. If the value is greater than the dynamic threshold, then multiple first failure probabilities are accepted; if the value is accepted, then multiple first failure probabilities are not accepted.

[0082] when Greater than the dynamic threshold or 3 times in a row Within 100ms, the decision is made to switch to the expert rule base, using the alarm priorities suggested in the pre-stored expert rule base within the nuclear power plant. If Less than or equal to dynamic threshold If the fault identification network model passes the detection, the first alarm priority output by the fault identification network model will continue to be used.

[0083] In some embodiments, the anomaly detection model, i.e., the variational autoencoder, can also be used to detect the model and obtain its reconstruction error, which can then be used as network operating parameters input to the detection network model.

[0084] By conducting real-time monitoring of the fault identification network model and the anomaly detection model, the system ensures seamless switching to the expert rule base when the model becomes unreliable. This effectively overcomes the risks of unreliability, uninterpretability, and overfitting that may occur with a single model under complex operating conditions in nuclear power plants, thus building a reliable defense and greatly improving the robustness, safety, and acceptability of nuclear power plant alarm priority outputs in practical engineering.

[0085] In some embodiments, reference Figure 5 The intelligent dynamic adjustment method for alarm priority also includes calculating the confidence level of the first alarm priority, specifically including the following steps S501 to S503.

[0086] In step S501, a first confidence level is calculated based on multiple first fault probabilities and the number of multiple fault types.

[0087] The first confidence level can be calculated according to the following formula (9). (9) in, The first confidence level is defined, ranging from [0, 1]. A value closer to 1 indicates greater certainty in the fault identification network model's calculation of the fault probability; a value closer to 0 indicates greater uncertainty. H(P) represents the entropy of multiple first fault probabilities P, measuring the uncertainty of these probabilities. A larger entropy value indicates a more dispersed distribution of the probabilities P, resulting in higher uncertainty; a smaller entropy value indicates a more concentrated distribution of the probabilities P, resulting in lower uncertainty. K represents the number of fault types.

[0088] In step S502, based on the deterministic function, according to the first security parameter Second safety parameter and third safety parameters Calculate the second confidence level.

[0089] Second confidence level It can be calculated using the following formula (10), (10) The deterministic function is calculated using the following formula (11): (11) In step S503, the confidence level of the first alarm priority is calculated based on the comprehensive confidence level, the first confidence level, the second confidence level, and the corresponding confidence level weights.

[0090] The confidence level of the first alarm priority is calculated using the following formula (12). : (12) in, This is used to adjust the contribution of the overall confidence level, the first confidence level, and the second confidence level to the overall confidence level based on their corresponding weights. It is typically set to... Recommended value The reason why the confidence level of the first alarm priority is based on a weighted geometric mean rather than an arithmetic mean is that the geometric mean is more sensitive to low values. The confidence level of the first alarm priority is dominated by the fault type with a low fault probability value, which can more conservatively reflect the true health status of the nuclear power plant system.

[0091] In some embodiments, based on the confidence level of the first alarm priority The confidence level and safety rule level are used to determine the alarm priority of a nuclear power plant.

[0092] Specifically, when the confidence level of the first alarm priority... If the confidence level is greater than 0.8 and the safety rule is Level 1, then the second alarm priority is output as the alarm priority for the nuclear power plant. When the safety rule is Level 2, the alarm priority for the nuclear power plant is calculated based on the first and second alarm priorities, as well as their corresponding first and second preset weights. The value of the first preset weight is increased; for example, the first and second preset weights can be 0.7 and 0.3, respectively. When the safety rule is Level 3, then the first alarm priority is output as the alarm priority for the nuclear power plant. Furthermore, the confidence level of this first alarm priority is... Within the range, dynamic threshold Use the base threshold .

[0093] When the confidence level of the first alarm priority If the alarm priority is less than or equal to 0.8 and greater than 0.6, and the safety rule is Level 1, then the second alarm priority is output as the alarm priority for the nuclear power plant. When the safety rule is Level 2, the alarm priority of the nuclear power plant is calculated based on the first and second alarm priorities, and their corresponding first and second preset weights, respectively. The value of the second preset weight is increased; for example, the first and second preset weights can be 0.3 and 0.7, respectively. When the safety rule is Level 3, the alarm priority of the nuclear power plant is calculated based on the first and second alarm priorities, and their corresponding first and second preset weights, respectively. For example, the first and second preset weights can be 0.5 and 0.5, respectively. Furthermore, the confidence level of this first alarm priority is... Within the range, dynamic threshold Use low threshold Exemplary It can be 0.7.

[0094] When the confidence level of the first alarm priority If the value is less than or equal to 0.6 and greater than 0.4, the second alarm priority will be output as the alarm priority for the nuclear power plant, and a low confidence level will be displayed on the system display interface of the nuclear power plant.

[0095] When the confidence level of the first alarm priority If the result is less than or equal to 0.4, the results of the first and second alarm priorities are bypassed, and the recommended level in the expert rule base is adopted.

[0096] A deep fusion mechanism for multi-source information was constructed by combining the comprehensive credibility obtained from data validity verification, the first confidence level obtained from the first failure probability calculation, the second confidence level obtained from nuclear safety assessment, and the safety rule level. Under the constraints of this mechanism, the nuclear power plant system strictly ensures the principle of prioritizing regulatory compliance, effectively suppressing false alarms and missed alarms that may be generated by a single data-driven model under conditions of operating condition drift or data anomalies. This ensures that the alarm priority output maintains high reliability and engineering practicality in the complex nuclear power plant environment, significantly improving the intelligent diagnostic system's adaptability to changes in operating status and its decision-making robustness.

[0097] The intelligent dynamic adjustment method for alarm priority in nuclear power plants proposed in this application can output alarm priorities according to different fault types, i.e. different operating conditions, and the output alarm priorities are dynamically adjusted in real time according to the above algorithm.

[0098] In some embodiments, reference Figure 6 It also includes displaying the obtained nuclear power plant fault types and the output alarm priorities of the nuclear power plant on the system interface, including the following steps S601 to S604: In step S601, the contribution of each parameter type in the real-time operating parameter set is calculated based on the real-time operating parameter set and the nuclear power plant baseline parameters.

[0099] The contribution of each parameter type in the real-time operating parameter set to the output nuclear power plant alarm priority can be calculated using an integrated gradient algorithm, according to the following formula (13): (13) The formula calculates the contribution of different parameter categories. This represents the integrated gradient value of the i-th parameter category in the real-time operating parameter set. It is a real number, which can be positive or negative. The larger its absolute value, the greater the contribution of that parameter category to the output alarm priority. Baseline parameters for nuclear power plants, For the i-th nuclear power plant baseline parameter, the selection rules for the nuclear power plant baseline parameter are as follows: all parameters that are zero, the average value of each operating parameter within the normal operating range, or the standard operating parameter value specified in the technical specifications, such as standard temperature (290°C) or standard pressure (15.0MPa), etc. These are the current real-time operating parameter values ​​of the nuclear power plant. For a certain type of real-time operating parameter, such as temperature (300°C) or pressure value (15.5MPa).

[0100] F(x) represents a layer in a fault identification network model, such as a layer in a Transformer model. These are interpolation parameters, ranging from [0,1], and their control is derived from the nuclear power plant baseline parameters. to real-time operating parameters The degree of interpolation between them, when The interpolation path is entirely within the nuclear power plant baseline parameter x', when The interpolation path is entirely based on the current real-time operating parameter value x of the nuclear power plant.

[0101] In step S602, the parameter types are sorted according to their contribution, and the top N parameter types are selected in descending order, where N is greater than or equal to 5 and less than the total number of parameter types.

[0102] In step S603, based on the fault type and alarm priority of the nuclear power plant, a retrieval algorithm is used to retrieve the corresponding fault handling measures from the knowledge base.

[0103] Based on semantic similarity and pattern matching, relevant fault handling measures are retrieved from the procedure knowledge base, which contains a network of associations between technical specifications, operating procedures, and historical cases.

[0104] In step S604, an explanation statement corresponding to the alarm priority of the nuclear power plant is generated based on the fault type, alarm priority, contribution of N parameter types, and fault handling measures.

[0105] Based on the interpretation template library, the system generates a fluent natural language statement by considering the fault type, alarm priority, contribution of parameter types, and fault handling measures, and displays it on the system interface. The template library contains over 200 predefined templates, covering various decision-making scenarios. This method forms the interpretation statement module.

[0106] In some embodiments, operators use an operator workstation to provide feedback on the output explanations, accepting or rejecting given alarm priorities and fault types. The collected feedback data is used to incrementally learn or batch retrain the fault identification network model. The operator workstation employs a dual-screen display design: the main screen displays alarm priorities, while the secondary screen displays fault types, the contribution of parameter types, and fault handling measures. This forms a continuous optimization loop of human-machine collaboration, significantly improving user trust, operational acceptability, and model self-evolution capabilities of the nuclear power intelligent auxiliary decision-making system.

[0107] In some embodiments, the anomaly detection model and fault identification network model are deployed on a dual-machine hot standby industrial server cluster to ensure high availability of critical models. The detection network model is deployed in a separate safety PLC, such as a Siemens S7-1500F, and the interpretation module runs on an AI server, interacting with models formed by other methods via a REST API. The system adopts a distributed deployment, allowing each model to run and be maintained independently, ensuring a safe switchover can still be performed in the event of a major model failure.

[0108] The method for determining alarm priority in nuclear power plants proposed in this application takes only 300ms to complete, which meets the real-time requirements of nuclear power plants and improves processing efficiency.

[0109] refer to Figure 7 To better explain the proposed method for determining the alarm priority of nuclear power plants, a specific embodiment is given below. It includes the following steps: Step S1: Data acquisition and preprocessing.

[0110] In this step, sensors acquire real-time operating parameters of the nuclear power plant, such as water level (45%, below the normal range of 50-60%), feedwater flow rate (10 kg / s, below the normal value of 15 kg / s), and steam flow rate (15 kg / s, within the normal range). Data preprocessing methods include data cleaning, such as removing noise and anomalies, and feature extraction, such as calculating the rate of change of water level and flow deviation. After the above processing, a standardized feature matrix of multiple real-time operating parameters is output.

[0111] The step is S2: Output the first alarm priority.

[0112] In this step, the real-time operating parameters are first validated to obtain a reliable set of real-time operating parameters. This set is then input into the fault identification network model to calculate multiple first fault probabilities. The calculation results are as follows: the probability of fault type "water supply system fault" is 0.85, the highest value; the probability of fault type "instrument false alarm" is 0.12, the second highest value. Next, a safety assessment is performed based on the multiple first fault probabilities, resulting in a first safety parameter of 0.75 (high risk), a second safety parameter of 0.30 (low risk), and a third safety parameter of 0.20 (low risk). The maximum value of these three is taken as the comprehensive safety parameter, which is 0.75. Based on the preset risk level strategy, this value corresponds to a high-risk level, thus determining the fault type as "water supply system fault," and its first alarm priority as high priority.

[0113] Step S3: Output the second alarm priority.

[0114] The real-time operating parameter set of the nuclear power plant is input into the Rete model. In the safety rule base, the safety rules are matched. If the safety rule level is the third level, the first alarm priority is output as the alarm priority of the nuclear power plant, that is, the high priority.

[0115] Step S4: Monitor the network model.

[0116] In this step, the network model used in step S2 is monitored, and the second fault probability is calculated. Its value is 0.2, which is within the normal range. The output of the first alarm priority and the subsequent output of the second alarm priority are allowed, without triggering the expert rule base.

[0117] Step S5: Generate interpretation statements.

[0118] In this step, the top-contributing real-time operating parameters are water level (45%), water supply flow rate (30%), and steam flow rate (15%). The retrieved procedure is "Water Supply System Fault Handling Procedure", the fault type is water supply system fault, the alarm priority is high priority, and the recommended operation is to start the backup water source and adjust the water supply control.

[0119] Step S6: Feedback and Optimization.

[0120] In this step, the operator has 5 seconds to confirm the alarm information, activate the backup water source, adjust the water supply control, and make a subjective evaluation, such as marking the above display result as the correct decision.

[0121] In some embodiments, referring to Table 1, the comparative example uses only the expert rule base to retrieve alarm priorities for nuclear power plants, while the embodiment uses the nuclear power plant alarm priority determination method proposed in this application. Through simulation testing of historical data, the embodiment shows a significant improvement over the comparative example in many indicators, demonstrating that the method proposed in this application significantly improves the effectiveness and efficiency of nuclear power plants in fault alarm handling.

[0122] Table 1

[0123] In another aspect, this application also proposes an electronic device including a memory and a processor, wherein the memory is used to store instructions executable by the processor, and the processor is used to execute the instructions to implement the dynamic adjustment method as described above.

[0124] refer to Figure 8 The schematic diagram of the electronic device shown illustrates that the electronic device 700 is used to implement the methods described above. The electronic device may include an internal communication bus 701, a processor 702, a read-only memory (ROM) 703, a random access memory (RAM) 704, a communication port 705, and a hard disk 706. The internal communication bus 701 enables data communication between the components of the electronic device 700. The processor 702 can perform judgments and issue prompts, and may include a CPU and a GPU. In some embodiments, the processor 702 may consist of one or more processors. The communication port 705 enables data communication between the electronic device 700 and external devices. In some embodiments, the electronic device 700 can send and receive information and data from a network through the communication port 705. The electronic device 700 may also include different forms of program storage units and data storage units, such as the hard disk 706, read-only memory (ROM) 703, and random access memory (RAM) 704, capable of storing various data files used for computer processing and / or communication, as well as possible program instructions executed by the processor 702. The processor executes these instructions to implement the main part of the method. The aforementioned intelligent dynamic adjustment method for alarm priority in nuclear power plants can be implemented as a computer program, stored in hard disk 706, and loaded into processor 702 for execution.

[0125] This application also includes a computer-readable medium storing computer program code that, when executed by a processor, implements the aforementioned intelligent dynamic adjustment method for alarm priorities in nuclear power plants.

[0126] Furthermore, this application uses specific terms to describe embodiments of the application. For example, "an embodiment," "one embodiment," and / or "some embodiments" refer to a particular feature, structure, or characteristic related to at least one embodiment of the application. Therefore, it should be emphasized and noted that "an embodiment," "one embodiment," or "an alternative embodiment" mentioned twice or more in different locations in this specification do not necessarily refer to the same embodiment. In addition, certain features, structures, or characteristics in one or more embodiments of the application can be appropriately combined.

[0127] Some aspects of this application can be executed entirely by hardware, entirely by software (including firmware, resident software, microcode, etc.), or by a combination of hardware and software. The aforementioned hardware or software may be referred to as a "data block," "module," "engine," "unit," "component," or "system." The processor may be one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DAPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), processors, controllers, microcontrollers, microprocessors, or combinations thereof. Furthermore, aspects of this application may manifest as computer products residing in one or more computer-readable media, including computer-readable program code. For example, computer-readable media may include, but are not limited to, magnetic storage devices (e.g., hard disks, floppy disks, magnetic tapes, etc.), optical discs (e.g., compressed CDs, digital multifunction DVDs, etc.), smart cards, and flash memory devices (e.g., cards, sticks, key drives, etc.).

[0128] A computer-readable medium may contain a propagated data signal containing computer program code, for example, on baseband or as part of a carrier wave. This propagated signal may take various forms, including electromagnetic, optical, and so on, or suitable combinations thereof. A computer-readable medium can be any computer-readable medium other than a computer-readable storage medium, which can be connected to an instruction execution system, apparatus, or device to enable communication, propagation, or transmission of a program for use. The program code located on the computer-readable medium can be propagated through any suitable medium, including radio, cable, fiber optic cable, radio frequency signals, or similar media, or any combination of the above media.

[0129] Similarly, it should be noted that, in order to simplify the description of the present application and thus aid in the understanding of one or more embodiments of the invention, the foregoing description of the embodiments of the present application sometimes combines multiple features into a single embodiment, drawing, or description thereof. However, this disclosure method does not imply that the subject matter of the application requires more features than those mentioned in the claims. In fact, the embodiments contain fewer features than all the features of the single embodiments disclosed above.

[0130] In some embodiments, numbers describing the quantity of components and attributes are used. It should be understood that such numbers used in the description of embodiments are modified in some examples with the terms "approximately," "approximately," or "generally." Unless otherwise stated, "approximately," "approximately," or "generally" indicates that the numbers are allowed to vary by ±20%. Accordingly, in some embodiments, the numerical parameters used in the specification and claims are approximate values, which may be changed depending on the characteristics required by individual embodiments. In some embodiments, numerical parameters should take into account specified significant digits and employ a general method of digit reservation. Although the numerical ranges and parameters used to confirm their breadth of scope in some embodiments of this application are approximate values, in specific embodiments, such values ​​are set as precisely as feasible.

[0131] Although this application has been described with reference to specific embodiments, those skilled in the art should recognize that the above embodiments are only used to illustrate this application, and various equivalent changes or substitutions can be made without departing from the spirit of this application. Therefore, any changes or modifications to the above embodiments within the essential spirit of this application will fall within the scope of the claims of this application.

Claims

1. A method for intelligent dynamic adjustment of alarm priority in nuclear power plants, characterized in that, Includes the following steps: Obtain multiple real-time operating parameters of the nuclear power plant; The data validity of the multiple real-time operating parameters is verified to obtain a set of real-time operating parameters; Based on the real-time operating parameter set, calculate multiple first fault probabilities corresponding to multiple fault types of the nuclear power plant; A first alarm priority is determined based on the plurality of first fault probabilities; Based on the real-time operating parameter set, match the security rules corresponding to the real-time operating parameter set in the security rule base, and obtain the second alarm priority based on the level of the security rule; The first alarm priority and the second alarm priority are processed according to a preset level strategy to obtain the alarm priority of the nuclear power plant.

2. The intelligent dynamic adjustment method for alarm priority in nuclear power plants as described in claim 1, characterized in that, The step of verifying the validity of the multiple real-time operating parameters and obtaining the real-time operating parameter set includes: The multiple real-time operating parameters are filtered to obtain multiple sets of reconfigurable real-time operating parameters corresponding to multiple subsystems in the nuclear power plant; Based on the multiple real-time operating parameters and the multiple sets of reconstruction real-time operating parameters, multiple reconstruction error values ​​are calculated; Based on the subsystem adjustment coefficients and the multiple reconstruction error values, obtain multiple credibility scores corresponding to the multiple subsystems respectively; The overall credibility is calculated based on the multiple credibility scores and the weights of the multiple subsystems. Based on the overall credibility and the preset credibility level, the set of real-time operating parameters is selected from the plurality of real-time operating parameters.

3. The intelligent dynamic adjustment method for alarm priority in nuclear power plants as described in claim 2, characterized in that, The step of determining the first alarm priority based on the plurality of first fault probabilities: Obtain the qualification parameters of the equipment in the nuclear power plant and the real-time operating status parameters of the nuclear power plant; The real-time operating status parameters, the multiple first fault probabilities, and the qualified parameters are respectively input into the first security network model, the second security network model, and the third security network model to obtain the first security parameter, the second security parameter, and the third security parameter; The maximum value among the first security parameter, the second security parameter, and the third security parameter is taken as the comprehensive security parameter; The comprehensive safety parameters are rated according to a preset risk level strategy to obtain the first alarm priority.

4. The intelligent dynamic adjustment method for alarm priority in nuclear power plants as described in claim 3, characterized in that, It also includes calculating the confidence level for the first alarm priority, which includes the following steps: A first confidence level is calculated based on the plurality of first fault probabilities and the number of the plurality of fault types; Based on the deterministic function, the second confidence level is calculated according to the first security parameter, the second security parameter, and the third security parameter; The confidence level of the first alarm priority is calculated based on the overall confidence level, the first confidence level, the second confidence level, and the corresponding confidence level weights.

5. The intelligent dynamic adjustment method for alarm priority in nuclear power plants as described in claim 1, characterized in that, The step of calculating multiple first fault probabilities corresponding to multiple fault types of the nuclear power plant based on the real-time operating parameter set includes: The real-time operating parameter set is input into the fault identification network model to calculate the probabilities of the multiple first faults. Obtain the operating parameters of the fault identification network model; A second fault probability is obtained based on the operating parameters of the fault identification network model. The second fault probability is the probability that the fault identification network model will fail within a preset time. The dynamic threshold is calculated based on the nuclear power plant's preset basic threshold, operational adjustment amount, and operating load; Determine whether the second fault probability is greater than the dynamic threshold. If the determination is no, then the plurality of first fault probabilities are accepted. If the determination is yes, then the plurality of first fault probabilities are not accepted.

6. The intelligent dynamic adjustment method for alarm priority in nuclear power plants as described in claim 1, characterized in that, Also includes: Output the explanation statement for the alarm priority of the nuclear power plant, which includes the following steps: Based on the real-time operating parameter set and the nuclear power plant baseline parameters, the contribution of each parameter type in the real-time operating parameter set is calculated; The parameter types are sorted by their contribution, and the top N parameter types are selected in descending order, wherein N is greater than or equal to 5 and less than the total number of parameter types; Based on the fault type and alarm priority of the nuclear power plant, a retrieval algorithm is used to retrieve the corresponding fault handling measures from the knowledge base; Based on the fault type, the alarm priority, the contribution of the N parameter types, and the fault handling measures, an explanation statement corresponding to the alarm priority of the nuclear power plant is generated.

7. The intelligent dynamic adjustment method for alarm priority in nuclear power plants as described in claim 1, characterized in that, The preset level strategy includes: When the safety rule is at level one, the second alarm priority shall be used as the alarm priority of the nuclear power plant. When the safety rule is level two, the alarm priority of the nuclear power plant is calculated based on the first alarm priority and the second alarm priority, as well as the corresponding first preset weight and second preset weight. When the safety rule is level three, the first alarm priority is used as the alarm priority of the nuclear power plant.

8. The intelligent dynamic adjustment method for alarm priority in nuclear power plants as described in claim 4, characterized in that, The preset level strategy includes: Determine the confidence interval in which the confidence level of the first alarm priority falls; The alarm priority of the nuclear power plant is determined based on the confidence interval of the first alarm priority and the level of the safety rule, wherein each confidence interval corresponds to multiple levels of the same safety rule.

9. An electronic device, characterized in that, include: Memory is used to store instructions that can be executed by the processor; as well as A processor for executing the instructions to implement the method as described in any one of claims 1-8.

10. A computer storage medium storing computer program code, characterized in that, The computer program code, when executed by a processor, implements the method as described in any one of claims 1-8.

Images