Nuclear power plant operation transient auxiliary decision method based on dynamic fusion decision

Abstract

This application proposes a transient auxiliary decision-making method for nuclear power plant operation based on dynamic fusion decision-making, comprising: collecting multiple operating parameters of the nuclear power plant through multiple sensors; inputting the multiple operating parameters into an operating transient diagnostic model to obtain multiple probabilities, wherein the multiple probabilities correspond one-to-one with multiple operating transient types; calculating a deterministic metric based on the multiple probabilities; generating at least one first operating decision and at least one corresponding first weight based on the deterministic metric, the multiple probabilities, and the multiple operating transient types; generating at least one second operating decision and at least one corresponding second weight based on the multiple operating parameters; calculating fusion weights corresponding to at least one first operating decision and at least one second operating decision based on the at least one first weight, at least one second weight, and the deterministic metric; and selecting at least one operating decision from at least one first operating decision and at least one second operating decision based on the fusion weights to obtain a fused operating decision.

Description

Technical Field

[0001] This application mainly relates to the field of nuclear power plants, and in particular to a transient auxiliary decision-making method for nuclear power plant operation based on dynamic fusion decision-making, electronic equipment, and computer storage medium. Background Technology

[0002] Against the backdrop of global energy structure transformation and the deep integration of artificial intelligence (AI) technology, nuclear power, as a clean, efficient, and stable baseload energy source, is ushering in unprecedented development opportunities. During the operation of nuclear power plants, quickly and accurately identifying transient operational conditions and making correct decisions is crucial to ensuring nuclear safety and operational efficiency. Currently, some decision-making support solutions exist, but all have significant shortcomings. Traditional threshold alarm systems can only alarm when a single parameter exceeds its limit, failing to understand the complex processes of multi-parameter coupling, resulting in high false alarm and false negative rates, and providing no decision-making suggestions. While some solutions incorporate machine learning models for fault diagnosis using single AI diagnostic models, these models are often "black boxes," lacking interpretability in their decision-making processes. Summary of the Invention

[0003] To address the aforementioned issues, this application proposes a dynamic fusion decision-making-based auxiliary decision-making method for nuclear power plant operation transients to improve the accuracy of transient type judgment.

[0004] In a first aspect, this application proposes a transient auxiliary decision-making method for nuclear power plant operation based on dynamic fusion decision-making, comprising the following steps: S1: collecting multiple operating parameters of the nuclear power plant through multiple sensors; S2: inputting the multiple operating parameters into an operating transient diagnostic model to obtain multiple probabilities, wherein the multiple probabilities correspond one-to-one with multiple operating transient types; S3: calculating a deterministic metric based on the multiple probabilities; S4: generating at least one first operating decision and at least one corresponding first weight based on the deterministic metric, the multiple probabilities, and the multiple operating transient types; S5: generating at least one second operating decision and at least one corresponding second weight based on the multiple operating parameters; S6: calculating fusion weights corresponding to the at least one first operating decision and the at least one second operating decision based on the at least one first weight, the at least one second weight, and the deterministic metric; S7: selecting at least one operating decision from the at least one first operating decision and the at least one second operating decision based on the fusion weights to obtain a fused operating decision.

[0005] 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.

[0006] 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.

[0007] Compared with the prior art, the beneficial effects of this application are as follows: (1) By constructing a transient diagnostic model, the transient type of operation is identified based on multiple operating parameters and the corresponding probability is generated, thereby effectively dealing with complex working conditions with multiple parameters coupled.

[0008] (2) By simultaneously using the operational transient type and its probability output by the operational transient diagnostic model to obtain the first operational decision and the second operational decision generated based on the operational parameters, the misjudgment caused by the uncertainty of a single decision source is effectively avoided, the overall accuracy and robustness of the operational transient type processing decision are significantly improved, and the problem of the uninterpretability of the operational transient diagnostic model itself is also alleviated. Attached Figure Description

[0009] 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 transient auxiliary decision-making method for nuclear power plant operation based on dynamic fusion decision-making provided in the embodiments of this application; Figure 2 This is a schematic diagram of the method for obtaining multiple probabilities by running a transient diagnostic model, as provided in an embodiment of this application. Figure 3 This is a schematic flowchart of a method for obtaining a first operational decision and a first weight provided in an embodiment of this application; Figure 4 This is a schematic flowchart of the method for obtaining the second operational decision and the second weight provided in the embodiments of this application; Figure 5 This is a schematic diagram of the method for obtaining fusion weights provided in an embodiment of this application; Figure 6 This is a schematic flowchart of a method for reducing the weight of preset rules provided in an embodiment of this application; Figure 7 This is a device block diagram of a transient auxiliary decision-making method for nuclear power plant operation based on dynamic fusion decision-making, provided in an embodiment of this application. Detailed Implementation

[0010] 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.

[0011] 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.

[0012] 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.

[0013] 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.

[0014] 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.

[0015] 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.

[0016] Flowcharts are used in this application to illustrate the operations performed by the system 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 steps may be removed from these processes.

[0017] The following specific embodiments illustrate the transient auxiliary decision-making method for nuclear power plant operation based on dynamic fusion decision-making of this application.

[0018] refer to Figure 1 This application proposes a transient auxiliary decision-making method for nuclear power plant operation based on dynamic fusion decision-making, including steps S110 to S170.

[0019] In step S110, multiple operating parameters of the nuclear power plant are collected in real time through multiple sensors.

[0020] Specifically, the nuclear power plant uses a Distributed Control System (DCS) to collect real-time readings from multiple sensors. These sensors include temperature, pressure, and flow sensors. The acquired readings serve as multiple operating parameters, such as temperature, pressure, flow rate, or power. The DCS processes these operating parameters along with other information from the sensors, encapsulating the processing results into a structure. This structure contains the following fields: acquisition time, original value, confidence flag (including suspicious and reliable flags), physical rule code, and parameter weight. The physical rule code is the physical rule code from the physical rule base triggered when the current operating parameter is assigned a suspicious flag. The parameter weight is calculated in real-time according to the physical rules after the current operating parameter has been verified. The confidence flag is the result of verification using physical rules. If the verification passes, the field in the confidence flag is reliable; if it fails, the field is suspicious.

[0021] In some embodiments, multiple operating parameters are validated according to a physical rule base to assign credibility tags to the multiple operating parameters. Specifically, it is determined whether the deviation of the multiple operating parameters calculated according to physical rules in the physical rule base is greater than the dynamic threshold corresponding to the physical rule. If the determination is no, multiple operating parameters are assigned credibility tags; if the determination is yes, multiple operating parameters are assigned suspicious tags, and a parameter anomaly warning is triggered.

[0022] The physics rule base contains multiple rules based on the actual operating conditions of the nuclear power plant and physical laws (such as the law of conservation of energy). For example, physics rule R1 is... in, The heat power carried away by the reactor core by the primary coolant. The heat power absorbed by the secondary working fluid (steam or feedwater) This refers to the heat lost from the primary circuit to the environment or insulation layer, This is the dynamic threshold. The heat output of the first loop should equal the heat absorbed by the second loop plus the unusable heat loss. If the deviation calculated using physical rule R1 is less than the dynamic threshold... If the calculated deviation is greater than or equal to the dynamic threshold, then a confidence marker is assigned to the current operating parameters; If a fault is detected, it indicates a possible primary loop leak (reasoned loss of reactor coolant leading to reduced heat removal) or damage to the steam generator heat transfer tubes. The current operating parameters will be flagged as suspicious, and a data quality warning will be triggered on the system interface, such as "Primary loop thermal power parameters may have measurement deviations, please be aware." Furthermore, the time of the event, the parameters involved, and the amount of deviation will be recorded in a dedicated "Data Anomaly Log" for maintenance personnel to troubleshoot sensor faults later.

[0023] In some embodiments, the start time is defined as the moment when the deviation first exceeds or equals a dynamic threshold, and the end time is defined as the period after which the deviation falls below the dynamic threshold and remains below it for a preset duration. All operating parameters between the start and end time points are marked as suspicious. More specifically, the nuclear power plant's system records the start time when the deviation first exceeds or equals the dynamic threshold and continuously monitors it. When the deviation falls below the dynamic threshold and remains below it for a preset duration (e.g., 10 seconds), the end time is recorded. All operating parameters between the start and end time points are marked as suspicious.

[0024] In some embodiments, the dynamic threshold is divided into a primary dynamic threshold and a secondary dynamic threshold. For example, the operating parameter currently used for judgment is the nuclear power plant's operating power. The primary dynamic threshold is twice the normal fluctuation range of the nuclear power plant's operating power. For instance, if the designed normal operating power of the nuclear power plant is 3000MW, and the steady-state fluctuation is ±30MW (1%), then the primary dynamic threshold might be set to ±60MW. The secondary dynamic threshold is five times the normal fluctuation range of the nuclear power plant's operating power, or reaches a certain absolute engineering value (e.g., greater than 10% of the nuclear power plant's rated operating power).

[0025] In some embodiments, parameter weights are calculated in real time based on dynamic thresholds. Taking triggering physical rule R1 as an example, when the deviation is less than the first-level dynamic threshold, the parameter weights... The value is 1.0. When the deviation is greater than or equal to the first-level dynamic threshold and less than or equal to the second-level dynamic threshold, the parameter weight is calculated according to formula (1). As an example, formula (1) is as follows: (1) Parameter weights are The deviation is F, and the first-level dynamic threshold is The second-level dynamic threshold is The parameter weights range from [0.3, 0.8]. When the deviation exceeds the secondary dynamic threshold, the parameter weights... The range is within [0, 0.3). Therefore, after the running parameters are validated, the corresponding parameter weights will be assigned values.

[0026] In step 120, multiple operating parameters are input into the transient diagnostic model to obtain multiple probabilities, which correspond one-to-one with multiple transient types.

[0027] In some embodiments, reference Figure 2 The transient diagnostic model is run by sequentially connecting a convolutional neural network layer, a long short-term memory network layer, and a fully connected network layer. Multiple running parameters are input into the transient diagnostic model to obtain multiple probabilities, including steps S210 to S230.

[0028] In step S210, multiple operating parameters are arranged according to operating parameter type and time dimension to form a parameter matrix, with each row of the parameter matrix corresponding to a different operating parameter type.

[0029] The rows of the parameter matrix represent the types of operating parameters, such as pressure vessel water level, steam generator outlet pressure, or steam generator outlet steam flow rate. The columns of the parameter matrix represent the time dimension, i.e., the time step.

[0030] In step S220, the parameter matrix is ​​input into the convolutional neural network layer, and the convolutional kernel of the convolutional neural network layer slides along the rows of the parameter matrix to extract the associated features of multiple running parameters as parameter features.

[0031] In this convolutional neural network layer, the convolutional kernel does not slide along the time dimension, but rather along the parameter type dimension to capture operational parameter features as parameter features. For example, the parameter types in the parameter matrix could be temperature, pressure, or flow rate. By sliding the convolutional kernel along the parameter type dimension, it can capture the correlation features among these three parameters. In this way, the convolutional neural network layer can learn local correlation patterns between multiple operational parameter types and extract cross-parameter type collaborative variation features.

[0032] In step S230, the parameter features are sequentially input into the long short-term memory network layer and the fully connected network layer to obtain multiple probabilities.

[0033] The parametric features undergo a feature reshaping operation to obtain reshaped parametric features. Obtaining reshaped parametric features is to adapt them to the parameter matrix requirements of the input Long Short-Term Memory (LSTM) network layer. For example, the feature matrix of the parametric features can have a shape of (c, t), where c is the number of feature channels generated after convolution, and t is the time step. The feature reshaping operation reshapes the feature matrix of the parametric features to (t, c).

[0034] The remodeling parameter features are input into a Long Short-Term Memory (LSTM) network layer, which, through a gating mechanism, learns the long-term temporal dependencies of these features. Fully connected layers map the parameter features output from the LSTM layer into multiple probabilities. These probabilities correspond one-to-one with multiple operational transient types. Nuclear power plant systems predetermine multiple operational transient types, and the number of these types is fixed. Thus, this hybrid spatial-temporal structure offers higher feature extraction efficiency and diagnostic accuracy. Furthermore, by constructing an operational transient diagnostic model, operational transient types are identified based on multiple operational parameters, and corresponding probabilities are generated, effectively addressing complex operating conditions involving multiple coupled parameters.

[0035] Continue to refer to Figure 1 In step 130, a deterministic metric is calculated based on multiple probabilities.

[0036] In some embodiments, a deterministic metric is calculated based on the entropy values ​​of multiple probabilities, and the deterministic metric can be calculated according to formula (2).

[0037] (2)

[0038] in, For deterministic measurement, , Entropy values ​​for multiple probabilities, , Let P be the probability of a certain transient type of operation, and let P be a probability vector formed by multiple probabilities. The number of transient types to run.

[0039] In some embodiments, multiple probability values ​​are sorted from largest to smallest, and the difference between the first probability and the second probability is calculated, with the difference used as a deterministic measure.

[0040] In step S140, at least one first operational decision and at least one corresponding first weight are generated based on deterministic metrics, multiple probabilities, and multiple operational transient types.

[0041] First, multiple fuzzy language descriptions are generated based on deterministic metrics, multiple probabilities, and multiple transient operation types. Then, at least one preset rule and at least one preset rule weight from the expert rule base are matched based on these multiple language descriptions. The preset rule and preset rule weight are converted into at least one first operation decision and at least one corresponding first weight.

[0042] In some embodiments, reference Figure 3The process of generating at least one first operational decision and at least one corresponding first weight based on deterministic metrics, multiple probabilities, and multiple operational transient types includes steps S310 to S320.

[0043] In step S310, the deterministic metric, multiple probabilities, and multiple runtime transient types are mapped to multiple fuzzy language descriptions, each of which includes a runtime transient type and a degree of urgency.

[0044] Specifically, deterministic metrics, multiple probabilities, and multiple transient operation types are mapped to at least one fuzzy language description according to mapping rules. An exemplary mapping rule is as follows: when the deterministic metric is greater than 0.8 and the probability is greater than 0.7, the urgency level is very high; when the deterministic metric is greater than 0.8 and the probability is less than or equal to 0.7 but greater than 0.5, the urgency level is high; when the deterministic metric is less than or equal to 0.8 and the probability is less than or equal to 0.5 but greater than 0.5, the urgency level is medium; when the deterministic metric is less than or equal to 0.8 and the probability is less than or equal to 0.5 but greater than 0.5, the urgency level is low; and when the deterministic metric is less than or equal to 0.8 and the probability is less than or equal to 0.3, the urgency level is low. For example, the probability of a water loss incident is 0.88, the deterministic measure is 0.82, and the mapping result is [Event Type: Water Loss Incident, Urgency Level: Very High].

[0045] In step S320, multiple preset rules and their corresponding preset rule weights from the expert rule base are matched according to multiple fuzzy language descriptions. Specifically, the expert rule base includes multiple IF-THEN rules, and the IF-THEN rules include multiple preset rules and their corresponding preset rule weights. For example, the IF-THEN rules can be: Rule R001: If the IF condition is event type = loss-of-coolant accident and urgency level = very high, then the THEN operation is emergency shutdown (1.0); Rule R002: If the IF condition is event type = loss-of-coolant accident and urgency level = very high, then the THEN operation is to start the emergency core cooling system (1.0); Rule R003: If the IF condition is event type = loss-of-coolant accident and urgency level = very high, then the THEN operation is to isolate the main steam system (0.9); Rule R004: If the IF condition is event type = loss-of-coolant accident and urgency level = high, then the THEN operation is to reduce power to 20% (0.8); Rule R005: If the IF condition is event type = loss-of-coolant accident and urgency level = high, then the THEN operation is to prepare for startup of the Emergency Core Cooling System (ECCS) (0.7); Rule R006: If the IF condition is event type = loss-of-coolant accident and urgency level = medium, then the THEN operation is to enter the fault diagnosis procedure (0.6). Each IF-THEN rule is called a preset rule. The number in parentheses within a preset rule represents its priority score, which is the weight of the preset rule. For example, if the fuzzy language description is [Event Type: Water Loss Accident, Urgency Level: Very High], then the multiple preset rules and their corresponding weights are: rules R001 to R003 and their corresponding priority scores of 1.0, 1.0, and 0.9.

[0046] In step S330, at least one first operation decision and at least one corresponding first weight are generated based on multiple preset rules and corresponding preset rule weights.

[0047] Specifically, the preset rules and their weights matched in the expert rule base are converted into a first operational decision and a first weight. For example, as illustrated in step S320, the preset rules are rules R001 to R003 and their corresponding priority scores. In the case of a loss-of-coolant accident, the first operational decisions are emergency shutdown, activation of the emergency core cooling system, and isolation of the main steam system, with corresponding first weights of 1.0, 1.0, and 0.9. This fuzzy language description matching of the expert rule base effectively alleviates the problem of the inherent uninterpretability of the transient diagnostic model.

[0048] Continue to refer to Figure 1 In step S150, at least one second operation decision and at least one corresponding second weight are generated based on multiple operating parameters.

[0049] In some embodiments, reference Figure 4 The process of generating at least one second operational decision and at least one corresponding second weight based on multiple operating parameters includes steps S410 to S420.

[0050] In step S410, historical cases are matched against a historical case library based on multiple operating parameters. Each historical case includes at least one disposal strategy and at least one corresponding strategy weight. Each historical case in the library includes scenario data and a disposal plan. The scenario data consists of historical curves of multiple operating parameters before and after a specific type of operational transient event. The disposal plan is the disposal strategy adopted by the operator at the time and subsequently verified as correct and effective; the disposal strategy includes corresponding strategy weights. Operating parameter curves are generated based on multiple operating parameters. Similarity is calculated using the Dynamic Time Warping (DTW) algorithm, and the current operating parameter curve is matched against the operating parameter curves in the historical case library to obtain historical cases consistent with the current state of the nuclear power plant. This algorithm can effectively align sequences of different time lengths. The specific steps for matching using the DTW algorithm are as follows: Step 1: Record the time series of multiple operating parameters within the first preset duration T as the query sequence. , , , }, where P is the number of runtime parameter types.

[0051] Step 2: Record the time series of multiple running parameters of historical cases in the historical case library within the second preset time period s as candidate sequences. , ,…, The first preset duration T and the second preset duration S can be the same or different.

[0052] Step 3: Calculate the weighted Euclidean distance as the weighted local distance according to formula (3): (3) in, Let be the weighted local distance between the i-th time step of the query sequence and the j-th time step of the candidate sequence. For the first The physical weights of each running parameter. Physical weight of each operating parameter Typically, physical weights are pre-calculated and set during the nuclear power plant system construction phase. These weights are determined based on sensitivity analysis of physical models or expert scoring methods. Sensitivity analysis of physical models involves simulating specific operational transients using high-fidelity thermal-hydraulic simulations (such as RELAP5 or TRACE), observing the magnitude or rate of change of each parameter, and quantifying the parameter's sensitivity to the transient type. Higher sensitivity results in a higher weight. For example, consider two sequences of operational parameters: water level and chilled tube temperature. Analysis shows that water level is three times more sensitive to loss-of-coolant accidents than temperature; therefore, the physical weight for water level can be set to 0.75, and the physical weight for chilled tube temperature to 0.25. A set of physical weight templates is pre-stored for each operational transient type within the nuclear power plant system. Expert scoring methods involve organizing domain experts (operations engineers, safety analysts) to score the importance of each parameter based on physical intuition and operational experience, and then normalizing the scores to obtain the physical weights. By introducing physical weights into the running parameters, the weighted similarity calculation more accurately reflects the essential characteristics of the running transient type, and can also more accurately retrieve similar running transient types from the historical case library.

[0053] Step 4: Initialize the cumulative distance matrix D, which has dimensions T×S, and let... , indicating the first position of the cumulative distance matrix The value is equal to the weighted local distance at that location. For the first line, The first row of the cumulative distance matrix The value at each position is equal to the cumulative distance of the previous position. Add the weighted local distance of the current position For the first column, , representing the first column of the cumulative distance matrix The value at each position is equal to the cumulative distance of the previous position plus the weighted local distance of the current position. For the remaining positions This represents the cumulative distance for other positions within the cumulative distance matrix D. Equal to the weighted local distance of the current position Add the minimum cumulative distance among its three adjacent positions to the left, above, and upper left.

[0054] Step 5: Set the cumulative distance matrix D(T,S) to the value of the candidate sequence. Weighted dynamic time-warped distance between the query sequence C and the query sequence C The smaller the distance, the more similar the two time series are.

[0055] Repeat steps 1 to 5 above. For each historical case in the historical case library, calculate the weighted dynamic time warp distance using each set of physical weight templates pre-stored in the nuclear power plant system (one set of weight templates corresponds to each type of operational transient). Select the globally minimum weighted dynamic time warp distance from it. The corresponding historical cases are used as the matching results.

[0056] In step S420, at least one second operational decision and at least one corresponding second weight are generated based on historical cases.

[0057] Specifically, historical cases include response strategies, each with a strategy weight. Each response strategy and its corresponding strategy weight are used as the second operational decision and its corresponding second weight, with at least one second operational decision and its corresponding second weight. For example, taking a loss-of-coolant accident as an example, the current trend of various operating parameters is as follows: in the past minute, the pressure vessel water level has continued to drop rapidly, the cold pipe section temperature has dropped slowly, and the containment pressure has begun to rise slightly. The matching historical case is historical case C-102 in the case library, whose response strategy is recorded as: "① Confirm water level drop > 10% / minute (0.9), ② Emergency shutdown (0.8), ③ Sequentially start ECCS pumps A and B (0.6)," where the values ​​in parentheses are strategy weights. Therefore, the response strategy and strategy weight of historical case C-102 are the second operational decision and its corresponding second weight.

[0058] Continue to refer to Figure 1 In step S160, based on at least one first weight, at least one second weight, and a deterministic metric, the fusion weights corresponding to at least one first operational decision and at least one second operational decision are calculated respectively.

[0059] In some embodiments, reference Figure 5 The steps for obtaining the fusion weights include steps S510 to S520.

[0060] In step S510, based on the deterministic measure, the first deterministic measure and the second deterministic measure corresponding to at least one first operation decision and at least one second operation decision are obtained respectively, wherein the first deterministic measure and the second deterministic measure are calculated according to formulas (4) and (5).

[0061] (4) (5) in For deterministic measurement, As the first deterministic measure, It is the second measure of certainty.

[0062] In step S520, the fusion weight is calculated based on the first deterministic metric, the second deterministic metric, at least one first weight and at least one second weight, wherein the fusion weight can be calculated according to formula (6).

[0063] (6) To integrate weights, As the first weight, As the second weight, As the first deterministic measure, This serves as the second deterministic measure. For a second operational decision that differs from the first operational decision, the second weight is indicated. The first weight is 0; for a first operational decision that differs from the second operational decision, the first weight indicates... The value is 0. For example, the deterministic metric is 0.8, and the multiple first operational decisions are emergency shutdown (1.0), activation of the emergency core cooling system (1.0), and isolation of the main steam system (0.9), with corresponding first weights of 1.0, 1.0, and 0.9, respectively. The multiple second operational decisions are confirmation of water level drop >10% / minute (0.9), emergency shutdown (0.8), and sequential activation of ECCS pumps in columns A and B (0.6), with corresponding second weights of 0.9, 0.8, and 0.6, respectively. The fusion weight of emergency shutdown is calculated according to formula (6). Confirmed water level drop >10% / minute The fusion weight for activating the emergency core cooling system is The fusion weight of the ECCS pumps in columns A and B is started sequentially. The fusion weight of the isolated main steam system is .

[0064] Continue to refer to Figure 1 In step S170, at least one operation decision is selected from at least one first operation decision and at least one second operation decision according to the fusion weight to obtain the fusion operation decision.

[0065] In some embodiments, the fusion weights are sorted from largest to smallest, and at least one operation decision is selected from at least one first operation decision and at least one second operation decision as the fusion operation decision based on the sorting result. In other words, when there is only one first operation decision and one second operation decision, and the operation decisions are the same, there is only one fusion operation decision; or the first operation decision and the second operation decision each include multiple operation decisions, which are sorted according to the fusion weights, and the top N of the sorting results are selected, where N is an integer and the value of N is in the range of [1,5]. For example, continuing to use the first and second operation decisions and the corresponding fusion weights from step S520, the fusion weights are sorted according to the fusion weights, and the sorting result is: emergency shutdown operation (0.96), start emergency core cooling system (0.8), isolate main steam system (0.72), confirm water level drop >10% / minute (0.18), start ECCS pumps A and B in sequence (0.12), where the weights in parentheses are the fusion weights calculated in step S520, and the above sorting result is the fusion operation decision. By simultaneously utilizing the first operational decision obtained through mapping between the transient diagnostic model and the expert rule base, and the second operational decision generated based on historical cases matched with operational parameters, misjudgments caused by the uncertainty of a single decision source can be effectively avoided, significantly improving the overall accuracy and robustness of decisions for handling transient operational types. Furthermore, utilizing operational decisions that have been validated in the historical case base provides a safety net for the entire decision-making system.

[0066] In some embodiments, reference Figure 6 The steps S610 to S640 involve reducing the weight of preset rules based on the operator's actions.

[0067] In step S610, the number of times each preset rule is matched by the fuzzy language description is recorded. Specifically, a recorder is maintained to record the number of times each preset rule is matched by the fuzzy language description in the expert rule base.

[0068] In step S620, the number of times the first operation decision generated based on each preset rule is rejected by the operator is recorded. In other words, the number of times the first operation decision generated based on the preset rules in step S610 is rejected by the operator as a fusion decision is recorded in the recorder.

[0069] In step S630, the rejection rate of each preset rule is calculated based on the number of matches and the number of rejections, and the rejection rate is calculated according to formula (7): (7) in, For preset rules, This is the number of rejections allowed under this preset rule. This represents the number of matches for the preset rule. This represents the rejection rate.

[0070] In step S640, when the rejection rate is greater than the rejection threshold, the weight of the preset rule corresponding to each preset rule is reduced according to the preset decay function.

[0071] Set a rejection threshold The rejection rate is 0.3 if the rejection rate is greater than the rejection threshold. Then, the updated preset rule weights are calculated based on the decay function, and the decay function can be calculated using formula (8).

[0072] (8)

[0073] Where k is the attenuation intensity coefficient, which can be taken as 2. For the updated preset rule weights, This sets the pre-defined rule weights before the update. Furthermore, a lower limit (e.g., 0.1) is set for the pre-defined rule weights; these weights must not fall below this value to prevent rules from being completely discarded. This mitigates the impact of frequently rejected pre-defined rules on future fusion decisions, thereby reducing the probability of generating rejected decisions again.

[0074] In some embodiments, the nuclear power plant system collects operating parameters associated with correctly confirmed fusion decisions made by operators within a preset time period, forming a small new dataset. The fully connected layer parameters are kept frozen, and the unfrozen portion of the operational transient diagnostic model is trained using the new dataset with an extremely low learning rate (e.g., 1 / 10 or 1 / 100 of the initial learning rate) to prevent significant modifications to existing knowledge. This process is performed periodically (e.g., monthly) in an offline environment, and after verification, the updated operational transient diagnostic model is deployed to the nuclear power plant system.

[0075] In some embodiments, the nuclear power plant's system interface displays a fusion of decision and determinism, including the contribution of the 3-5 operating parameters and their responses that contribute most to the diagnosis of the current transient type. Furthermore, when the determinism metric is greater than 0.8, the nuclear power plant's system interface displays the specific transient type, such as a loss-of-coolant accident; when the determinism metric is less than or equal to 0.8, the nuclear power plant's system interface displays that the transient type cannot be determined.

[0076] 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 decision support method as described above.

[0077] refer to Figure 7The 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) 707, 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) 707, 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 decision support method can be implemented as a computer program, stored in hard disk 706, and loaded into processor 702 for execution.

[0078] This application also includes a computer-readable medium storing computer program code that, when executed by a processor, implements the aforementioned decision support method.

[0079] 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.

[0080] 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.).

[0081] 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.

[0082] 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.

[0083] 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.

[0084] 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 transient auxiliary decision-making method for nuclear power plant operation based on dynamic fusion decision-making, comprising the following steps: S1: Collect multiple operating parameters of the nuclear power plant through multiple sensors; S2: Input the multiple operating parameters into the transient diagnostic model to obtain multiple probabilities, which correspond one-to-one with multiple transient types; S3: Calculate a deterministic measure based on the plurality of probabilities; S4: Based on the deterministic metric, the plurality of probabilities, and the plurality of operational transient types, generate at least one first operational decision and at least one corresponding first weight; S5: Based on the multiple operating parameters, generate at least one second operation decision and at least one corresponding second weight; S6: Calculate the fusion weights corresponding to the at least one first operational decision and the at least one second operational decision based on the at least one first weight, the at least one second weight, and the deterministic metric; S7: Select at least one operation decision from the at least one first operation decision and the at least one second operation decision according to the fusion weight, and obtain a fused operation decision.

2. The transient auxiliary decision-making method for nuclear power plant operation based on dynamic fusion decision-making as described in claim 1, characterized in that, Before step S2, the method further includes: verifying the plurality of operating parameters according to the physical rule base, so as to assign a confidence label to the plurality of operating parameters.

3. The transient auxiliary decision-making method for nuclear power plant operation based on dynamic fusion decision-making as described in claim 2, characterized in that, The step of verifying the multiple operating parameters according to the physical rule base and assigning a confidence label to the multiple operating parameters includes: If the deviation of the multiple operating parameters calculated according to the physical rules in the physical rule base is greater than the dynamic threshold corresponding to the physical rule, then if the deviation is not greater, then the multiple operating parameters are assigned a trustworthy label; if the deviation is greater, then the multiple operating parameters are assigned a suspicious label and a parameter anomaly warning is triggered.

4. The transient auxiliary decision-making method for nuclear power plant operation based on dynamic fusion decision-making as described in claim 1, characterized in that, The transient diagnostic model includes a convolutional neural network layer, a long short-term memory network layer, and a fully connected network layer connected in sequence. Step S2 includes the following steps: The multiple operating parameters are arranged according to the operating parameter type and time dimension to form a parameter matrix, wherein each row of the parameter matrix corresponds to a different operating parameter type; The parameter matrix is ​​input into the convolutional neural network layer, and the convolutional kernel of the convolutional neural network layer slides along the rows of the parameter matrix to extract the correlation features of the multiple running parameters as parameter features; The parameter features are sequentially input into the long short-term memory network layer and the fully connected network layer to obtain the multiple probabilities.

5. The transient auxiliary decision-making method for nuclear power plant operation based on dynamic fusion decision-making as described in claim 1, characterized in that, Step S4 includes the following steps: The deterministic metric, the plurality of probabilities, and the plurality of operational transient types are mapped to a plurality of fuzzy language descriptions, each fuzzy language description including an operational transient type and a degree of urgency; Based on the multiple fuzzy language descriptions, match multiple preset rules and corresponding preset rule weights in the expert rule base; Based on the plurality of preset rules and the corresponding plurality of preset rule weights, at least one first operation decision and at least one corresponding first weight are generated.

6. The transient auxiliary decision-making method for nuclear power plant operation based on dynamic fusion decision-making as described in claim 1, characterized in that, Step S5 includes the following steps: Based on the multiple operating parameters, historical cases are matched in the historical case library, wherein the historical cases include at least one disposal strategy and at least one corresponding strategy weight; The at least one second operational decision and the corresponding at least one second weight are generated based on the historical cases.

7. The transient auxiliary decision-making method for nuclear power plant operation based on dynamic fusion decision-making as described in claim 1, characterized in that, Step S6 includes the following steps: Based on the deterministic measure, the first deterministic measure and the second deterministic measure corresponding to the at least one first operational decision and the at least one second operational decision are obtained respectively; The fusion weight is calculated based on the first deterministic metric, the second deterministic metric, the at least one first weight, and the at least one second weight.

8. The transient auxiliary decision-making method for nuclear power plant operation based on dynamic fusion decision-making as described in claim 1, characterized in that, Step S7 includes the following steps: Sort the fusion weights from largest to smallest; Based on the sorting results, at least one operation decision is selected from the at least one first operation decision and the at least one second operation decision as the fusion operation decision.

9. The transient auxiliary decision-making method for nuclear power plant operation based on dynamic fusion decision-making as described in claim 5, characterized in that, Also includes: Record the number of times each of the preset rules is matched by the fuzzy language description; Record the number of times the first operation decision generated based on each of the preset rules is rejected by the operator; The rejection rate for each preset rule is calculated based on the number of matches and the number of rejections. When the rejection rate is greater than the rejection threshold, the weight of the preset rule corresponding to each preset rule is reduced according to the preset decay function.

10. 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 1-9.

11. 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 1-9.

Images