Fault analysis method and device, computer device, storage medium and program product

Abstract

The application relates to a fault analysis method and device, computer equipment, a storage medium and a computer program product. The method comprises the following steps: determining target fault coding information of a target nuclear power station according to first operation parameters of the target nuclear power station and a target fault diagnosis model; in the case that the target fault coding information exists, performing similarity analysis on the first operation parameters based on second operation parameters to obtain a first similarity of the first operation parameters; performing similarity analysis on the first operation parameters based on third operation parameters to obtain a second similarity of the first operation parameters; determining first target operation parameters and second target operation parameters from the first operation parameters according to a first similarity threshold, a second similarity threshold, the first similarity and the second similarity, and displaying the first target operation parameters and the second target operation parameters. The fault analysis method provided by the application can effectively improve the work efficiency of the fault analysis task.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of nuclear power plant safety, and in particular to a fault analysis method and device, computer equipment, a storage medium and a computer program product. BACKGROUND

[0002] With the rapid development of nuclear power technology, people also pay more and more attention to the safety of nuclear power plants.

[0003] In the prior art, the fault diagnosis of a nuclear power plant is usually realized based on artificial intelligence. This method trains a model based on massive fault data to improve the accuracy of fault diagnosis.

[0004] However, on the one hand, the fault data of a nuclear power plant is usually small, resulting in poor accuracy of the fault diagnosis result obtained by using this method, and on the other hand, the fault diagnosis result obtained by this method has poor interpretability, and further analysis of the fault by technical personnel is required, thereby resulting in poor work efficiency of the nuclear power plant fault analysis task. SUMMARY

[0005] Therefore, it is necessary to provide a fault analysis method, device, computer equipment, computer readable storage medium and computer program product with high work efficiency to solve the above technical problems.

[0006] In a first aspect, the present application provides a fault analysis method, comprising:

[0007] determining target fault code information of a target nuclear power plant based on a first operation parameter of the target nuclear power plant and a target fault diagnosis model, the target fault code information being used to indicate a fault type of the target nuclear power plant; in the case where the target fault code information exists, performing similarity analysis on the first operation parameter based on a second operation parameter to obtain a first similarity of the first operation parameter, the second operation parameter being an operation parameter of the target nuclear power plant in a normal state; performing similarity analysis on the first operation parameter based on a third operation parameter to obtain a second similarity of the first operation parameter, the third operation parameter being an operation parameter with the same fault code information as the target fault code information; determining a first target operation parameter and a second target operation parameter according to a first similarity threshold, a second similarity threshold, the first similarity and the second similarity, and displaying the first target operation parameter and the second target operation parameter.

[0008] In one of the embodiments, the target fault coding information of the target nuclear power plant is determined according to the first operation parameter of the target nuclear power plant and the target fault diagnosis model, which includes: performing noise reduction processing on the first operation parameter based on a limiting value filtering algorithm and a Kalman filtering algorithm; performing feature extraction on the first operation parameter after the noise reduction processing based on a piecewise linear representation method to obtain shape features of the first operation parameter; and inputting the shape features into the target fault diagnosis model to determine the target fault coding information of the target nuclear power plant.

[0009] In one of the embodiments, the similarity analysis is performed on the first operation parameter based on the second operation parameter to obtain a first similarity of the first operation parameter, which includes: determining the similarity of the first operation parameter and the second operation parameter based on a dynamic time warping method, and determining the similarity as the first similarity of the first operation parameter.

[0010] In one of the embodiments, the similarity analysis is performed on the first operation parameter based on the third operation parameter to obtain a second similarity of the first operation parameter, which includes: determining a fault type of the target nuclear power plant according to the target fault coding information; determining the third operation parameter according to the fault type of the target nuclear power plant; determining the similarity of the first operation parameter and the third operation parameter based on a pattern distance method, and determining the similarity as the second similarity of the first operation parameter.

[0011] In one of the embodiments, the training method of the target fault diagnosis model includes: obtaining a training data set, the training data set including a plurality of labeled normal state operation parameters and labeled fault state operation parameters, the normal state operation parameters being operation parameters obtained when the target nuclear power plant is in a normal operation state, and the fault state operation parameters being operation parameters obtained when the target nuclear power plant is in a fault state; and determining the target fault diagnosis model according to the training data set and an initial fault diagnosis model.

[0012] In one of the embodiments, the target fault diagnosis model is determined according to the training data set and the initial fault diagnosis model, which includes: determining a small sample data set according to the training data set, and performing small sample learning training based on the small sample data set and the initial fault diagnosis model to obtain a first fault diagnosis model; and performing update optimization processing on the first fault diagnosis model based on a cross-entropy loss function algorithm and a momentum optimization algorithm to obtain a target fault diagnosis model.

[0013] In a second aspect, the present application further provides a fault device, which includes:

[0014] determining a target fault code information of the target nuclear power plant according to the first operation parameter and a target fault diagnosis model, the target fault code information being used to indicate a fault type of the target nuclear power plant;

[0015] performing similarity analysis on the first operation parameter based on a second operation parameter to obtain a first similarity of the first operation parameter in a case where the target fault code information exists, the second operation parameter being an operation parameter of the target nuclear power plant in a normal state;

[0016] performing similarity analysis on the first operation parameter based on a third operation parameter to obtain a second similarity of the first operation parameter, the third operation parameter being an operation parameter of which the fault code information is same as the target fault code information;

[0017] determining a first target operation parameter and a second target operation parameter according to a first similarity threshold, a second similarity threshold, the first similarity and the second similarity, and displaying the first target operation parameter and the second target operation parameter.

[0018] In an embodiment, the determining module is specifically configured to perform noise reduction processing on the first operation parameter based on a clipping value filtering algorithm and a Kalman filtering algorithm, perform feature extraction on the first operation parameter after the noise reduction processing based on a piecewise linear representation method to obtain a shape feature of the first operation parameter, and input the shape feature into the target fault diagnosis model to determine the target fault code information of the target nuclear power plant.

[0019] In an embodiment, the first executing module is specifically configured to determine the similarity of the first operation parameter and the second operation parameter based on a dynamic time warping method, and determine the similarity as the first similarity of the first operation parameter.

[0020] In an embodiment, the second executing module is specifically configured to determine the fault type of the target nuclear power plant according to the target fault code information, determine the third operation parameter according to the fault type of the target nuclear power plant, determine the similarity of the first operation parameter and the third operation parameter based on a pattern distance method, and determine the similarity as the second similarity of the first operation parameter.

[0021] In an embodiment, the fault analysis apparatus further comprises a fourth execution module configured to obtain a training data set, the training data set comprising a plurality of labeled normal state operation parameters and labeled fault state operation parameters, the normal state operation parameters being operation parameters obtained when the target nuclear power plant is in a normal operation state, and the fault state operation parameters being operation parameters obtained when the target nuclear power plant is in a fault state; and determine the target fault diagnosis model based on the training data set and an initial fault diagnosis model.

[0022] In an embodiment, the fourth execution module is specifically configured to determine a small sample data set based on the training data set, and perform small sample learning training based on the small sample data set and the initial fault diagnosis model to obtain a first fault diagnosis model; and perform update optimization processing on the first fault diagnosis model based on a cross-entropy loss function algorithm and a momentum optimization algorithm to obtain the target fault diagnosis model.

[0023] In a third aspect, the present application also provides a computer device. The computer device comprises a memory and a processor, the memory stores a computer program, and the processor implements the steps of any one of the first aspect when executing the computer program.

[0024] In a fourth aspect, the present application also provides a computer readable storage medium. The computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the steps of any one of the first aspect.

[0025] In a fifth aspect, the present application also provides a computer program product. The computer program product comprises a computer program, and the computer program is executed by a processor to implement the steps of any one of the first aspect.

[0026] The fault analysis method, device, computer device, storage medium and computer program product determine target fault coding information of the target nuclear power plant according to the first operation parameter of the target nuclear power plant and a target fault diagnosis model, the target fault coding information is used to indicate the fault type of the target nuclear power plant; in the case that the target fault coding information exists, the similarity of the first operation parameter is analyzed based on a second operation parameter to obtain a first similarity of the first operation parameter, the second operation parameter is an operation parameter of the target nuclear power plant in a normal state; the similarity of the first operation parameter is analyzed based on a third operation parameter to obtain a second similarity of the first operation parameter, the third operation parameter is an operation parameter with the same fault coding information as the target fault coding information; the first target operation parameter and the second target operation parameter are determined according to a first similarity threshold, a second similarity threshold, the first similarity and the second similarity, and the first target operation parameter and the second target operation parameter are displayed. The fault analysis method provided in the present application can not only diagnose the fault type of the nuclear power plant based on the operation parameter of the nuclear power plant, but also output the operation parameter corresponding to the fault state when the nuclear power plant is in a fault state, so as to facilitate the technical personnel to determine the corresponding fault equipment and the specific fault reason according to the operation parameter. The fault analysis method provided in the present application can effectively improve the explainability of the fault analysis result, and can further assist the technical personnel to analyze the fault of the nuclear power plant, thereby effectively improving the work efficiency of the nuclear power plant abnormal analysis task. BRIEF DESCRIPTION OF DRAWINGS

[0027] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the drawings needed to be used in the embodiments or the related art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0028] Figure 1 A flowchart of a fault analysis method in an embodiment;

[0029] Figure 2 A flowchart of a method for determining target fault coding information of a target nuclear power plant according to a first operation parameter of the target nuclear power plant and a target fault diagnosis model in an embodiment;

[0030] Figure 3 A flowchart of a method for analyzing the similarity of the first operation parameter based on a third operation parameter to obtain a second similarity of the first operation parameter in an embodiment;

[0031] Figure 4A flowchart of a training method of a target fault diagnosis model in an embodiment is shown in FIG. 1.

[0032] Figure 5 A flowchart of a method of determining the target fault diagnosis model according to the training data set and the initial fault diagnosis model in an embodiment is shown in FIG. 2.

[0033] Figure 6 A flowchart of a fault analysis method in another embodiment is shown in FIG. 3.

[0034] Figure 7 A block diagram of a fault analysis device in an embodiment is shown in FIG. 4.

[0035] Figure 8 A block diagram of a fault analysis device in another embodiment is shown in FIG. 5.

[0036] Figure 9 An internal structure diagram of a computer device in an embodiment is shown in FIG. 6. DETAILED DESCRIPTION

[0037] In order to make the purposes, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and should not be used to limit the present application.

[0038] With the rapid development of nuclear power technology, people also pay more and more attention to the safety of nuclear power plants.

[0039] In the prior art, the fault diagnosis of a nuclear power plant is usually realized based on artificial intelligence. This method trains a model based on massive fault data to improve the accuracy of fault diagnosis.

[0040] However, on the one hand, the fault data of a nuclear power plant is usually small, resulting in poor accuracy of the fault diagnosis result obtained by using this method, and on the other hand, the fault diagnosis result obtained by using this method has poor interpretability, and further analysis of the fault by technical personnel is required, thereby resulting in poor work efficiency of the fault analysis task of the nuclear power plant.

[0041] Therefore, the present application provides a fault analysis method which can effectively improve the work efficiency of the fault analysis task of a nuclear power plant.

[0042] The fault analysis method provided by the embodiments of the present application can be executed by a computer device, which can be a server.

[0043] In an embodiment, as shown in FIG. 1, a fault analysis method is provided, which comprises the following steps: Figure 1

[0044] ​In step 101, target fault coding information of the target nuclear power plant is determined according to the first operation parameter of the target nuclear power plant and a target fault diagnosis model.

[0045] The target fault coding information is used to indicate a fault type of the target nuclear power plant.

[0046] Optionally, the target nuclear power plant is a nuclear power plant that has a fault analysis requirement, and the first operation parameter can include reactor nuclear power, reactor thermal power, feedwater temperature, feedwater pressure, feedwater flow, steam temperature, steam pressure, steam flow, nuclear power positive change rate, and nuclear power negative change rate, and can further include other operation parameters of the target nuclear power plant.

[0047] Optionally, the target fault diagnosis model is used to output the target fault coding information of the target nuclear power plant according to the first operation parameter of the target nuclear power plant.

[0048] In one possible implementation, the target fault diagnosis model can be composed of a cascaded target Transformer model and a plurality of target fully connected neural networks. It should be noted that the number of neural network settings in the target fault diagnosis model can be determined according to the operation parameters that need to be evaluated. Based on repeated tests of the first operation parameter, the target fault diagnosis model determined by the present inventors is composed of a cascaded target Transformer model and two target fully connected neural networks.

[0049] In another possible implementation, the target fault diagnosis model can also be composed of only fully connected neural networks, or composed of only convolutional neural networks, or composed of only recurrent neural networks, or composed of a combination of the above neural networks.

[0050] As described above, the target fault coding information of the target nuclear power plant can be determined according to the first operation parameter of the target nuclear power plant and the target fault diagnosis model.

[0051] In one possible implementation, a table can be correspondingly set, in which a plurality of fault codes and a plurality of fault types corresponding to the fault codes are recorded. Based on the table and the target fault coding information, the fault type of the target nuclear power plant can be determined.

[0052] In another possible implementation, a fault information database can be set, in which a plurality of fault codes and a plurality of fault types corresponding to the fault codes are stored. Based on the target fault coding information, the fault information database can be queried to determine the fault type of the target nuclear power plant.

[0053] In step 102, similarity analysis is performed on the first operation parameter based on a second operation parameter to obtain a first similarity of the first operation parameter, in a case where target fault code information exists.

[0054] The second operation parameter is an operation parameter of the target nuclear power plant in a normal state.

[0055] In one possible implementation, the target fault diagnosis model outputs the target fault code information only in a case where the target nuclear power plant is in a fault state, and thus the target nuclear power plant is determined to be in the fault state only in a case where the target fault code information exists.

[0056] In another possible implementation, the target fault diagnosis model outputs the target fault code information regardless of whether the target nuclear power plant is in a fault state, and the target nuclear power plant is determined to be in the fault state based on the target fault code information, and then the fault type of the target nuclear power plant is analyzed in a case where the target nuclear power plant is in the fault state.

[0057] In an optional embodiment of the present application, the similarity analysis performed on the first operation parameter based on the second operation parameter to obtain the first similarity of the first operation parameter includes: determining the similarity of the first operation parameter and the second operation parameter based on a dynamic time warping method, and determining the similarity as the first similarity of the first operation parameter.

[0058] Optionally, the dynamic time warping method can be used to calculate the similarity of two time series. Specifically, a time correspondence relationship between a test template and a reference template is described based on a time warping function that meets a preset condition, to determine the similarity of the test template and the reference template.

[0059] In a possible implementation, if the target fault coding information exists, it is determined that the target nuclear power plant is in a fault condition, and it is determined that there is a fault parameter in the first operation parameter, the fault parameter being an operation parameter generated by a device of the target nuclear power plant in a fault working state, the first operation parameter being the test template described above, and the second operation parameter being the reference template described above. For example, the first operation parameter includes A1, B1 and C1 parameters, and the second operation parameter corresponds to A2, B2 and C2 parameters, where the A1 parameter is an operation parameter of an A device in the target nuclear power plant, the B1 parameter is an operation parameter of a B device in the target nuclear power plant, the C1 parameter is an operation parameter of a C device in the target nuclear power plant, the A2 parameter is an operation parameter of the A device in the target nuclear power plant in a normal working state, the B2 parameter is an operation parameter of the B device in the target nuclear power plant in a normal working state, and the C2 parameter is an operation parameter of the C device in the target nuclear power plant in a normal working state. The similarity of the A1 parameter and the A2 parameter, the similarity of the B1 parameter and the B2 parameter, and the similarity of the C1 parameter and the C2 parameter are determined respectively.

[0060] In an optional embodiment of the present application, the similarity of the first operation parameter and the third operation parameter can also be determined based on a mode distance method, and the similarity is determined as the similarity of the first operation parameter.

[0061] In another optional embodiment of the present application, the similarity of the first operation parameter and the second operation parameter can also be determined based on a shape distance algorithm, and the similarity is determined as the similarity of the first operation parameter.

[0062] Step 103: performing similarity analysis on the first operation parameter based on the third operation parameter to obtain a second similarity of the first operation parameter.

[0063] The third operation parameter is an operation parameter with the same fault coding information as the target fault coding information.

[0064] In a possible implementation, the target fault coding information is used to represent the fault type of the target nuclear power plant, but in actual operation, the same fault type may not result in the same fault cause or device, that is, the target fault coding can only determine the fault type, and each fault type corresponds to multiple fault causes. Therefore, after determining the fault type based on the target fault coding information, the fault cause needs to be determined again, and therefore, the third operation parameter needs to be obtained from the fault parameter database based on the fault type, and the second similarity of the first operation parameter is determined based on the third operation parameter.

[0065] Step 104, determining the first target operation parameter and the second target operation parameter according to the first similarity threshold, the second similarity threshold, the first similarity and the second similarity, and displaying the first target operation parameter and the second target operation parameter.

[0066] Optionally, the first similarity threshold can be set by the technician in advance.

[0067] In one possible implementation, the operation parameter of which the first similarity is less than the first similarity threshold in the first operation parameters can be determined as the first target operation parameter.

[0068] Optionally, as described above, after determining the first target operation parameter, the first target operation parameter also needs to be displayed, and the technician can determine the equipment in the fault state in the target nuclear power plant according to the displayed first target operation parameter.

[0069] In one possible implementation, if the first target operation parameter determined according to the first similarity threshold and the first similarity of the first operation parameter is the C1 parameter, that is, the C1 parameter has a large difference with the C1 parameter obtained in the normal working state, the C1 parameter is displayed, and the technician can determine that the C equipment in the target nuclear power plant can be in a fault state according to the displayed C1 parameter, and perform fault checking or fault analysis on the C equipment.

[0070] In another possible implementation, if the first target operation parameter determined according to the first similarity threshold and the first similarity of the first operation parameter is the A1 parameter and the B1 parameter, that is, the A1 parameter and the B1 parameter have a large difference with the A1 parameter and the B1 parameter obtained in the normal working state, the A1 parameter and the B1 parameter are displayed, and the technician can determine that the A equipment and the B equipment in the target nuclear power plant can be in a fault state according to the displayed A1 parameter and the B1 parameter, and perform fault checking or fault analysis on the A equipment and the B equipment.

[0071] Optionally, the second similarity threshold can be set by the technician in advance.

[0072] In one possible implementation, the operation parameter of which the second similarity is greater than the second similarity threshold can be determined as the second target operation parameter.

[0073] Optionally, as described above, after determining the second target operation parameter, the second target operation parameter also needs to be displayed, and the technician can determine the specific fault reason of the target nuclear power plant according to the displayed second target operation parameter.

[0074] The fault analysis method determines target fault coding information of the target nuclear power plant according to the first operation parameter of the target nuclear power plant and a target fault diagnosis model, the target fault coding information is used to indicate a fault type of the target nuclear power plant; in the case that the target fault coding information exists, similarity analysis is performed on the first operation parameter based on a second operation parameter to obtain a first similarity of the first operation parameter, the second operation parameter is an operation parameter of the target nuclear power plant in a normal state; similarity analysis is performed on the first operation parameter based on a third operation parameter to obtain a second similarity of the first operation parameter, the third operation parameter is an operation parameter with the same fault coding information as the target fault coding information; first target operation parameters and second target operation parameters are determined according to a first similarity threshold, a second similarity threshold, the first similarity and the second similarity, and the first target operation parameters and the second target operation parameters are displayed. By using the fault analysis method provided in the application, not only can the fault type of the nuclear power plant be diagnosed based on the operation parameter of the nuclear power plant, but also the operation parameter corresponding to the fault state can be output when the nuclear power plant is in a fault state, which facilitates the technical personnel to determine the corresponding fault equipment and the specific fault reason according to the operation parameter. By using the fault analysis method provided in the application, the explainability of the fault analysis result can be effectively improved, and the technical personnel can be assisted to further analyze the fault of the nuclear power plant, thereby effectively improving the work efficiency of the nuclear power plant abnormality analysis task.

[0075] In one embodiment, as shown in Figure 2 The target fault coding information of the target nuclear power plant is determined according to the first operation parameter of the target nuclear power plant and a target fault diagnosis model, including the following steps:

[0076] Step 201, performing noise reduction processing on the first operation parameter based on a limiting value filtering algorithm and a Kalman filtering algorithm.

[0077] In an optional embodiment of the application, the first operation parameter of the target nuclear power plant can be obtained first.

[0078] In one possible implementation, the technical personnel can determine a target device from a plurality of devices in the target nuclear power plant according to actual needs, and obtain the operation parameter of the target device, and the operation parameter of the target device is determined as the first operation parameter.

[0079] In another possible implementation, the plurality of devices in the target nuclear power plant can also be determined as target devices, and the operation parameter of the target device is obtained, and the operation parameter of the target device is determined as the first operation parameter.

[0080] Optionally, the amplitude limiting value filtering algorithm is to set a suitable limiting value, and when the amplitude value of a sampling point increases or decreases by a value exceeding the limiting value relative to the previous sampling point and the next sampling point, the point is considered to be a pulse point, and the average value of the previous two sampling points is used as the result of this filtering.

[0081] Optionally, the Kalman filtering algorithm refers to a state estimation algorithm combining prior experience and measurement update.

[0082] In a possible implementation manner, the first operating parameter can be denoised only based on the amplitude limiting value filtering algorithm.

[0083] In another possible implementation manner, the first operating parameter can also be denoised only based on the Kalman filtering algorithm.

[0084] In another possible implementation manner, the first operating parameter can be first denoised based on the amplitude limiting value filtering algorithm, and then denoised based on the Kalman filtering algorithm.

[0085] Step 202, performing feature extraction on the denoised first operating parameter based on a piecewise linear representation method to obtain a shape feature of the first operating parameter.

[0086] Optionally, the piecewise linear representation method is a time series data mining method.

[0087] In a possible implementation manner, the shape feature of the first operating parameter can be obtained by performing feature extraction on the denoised first operating parameter based on the piecewise linear representation method.

[0088] In another possible implementation manner, the shape feature of the first operating parameter can also be obtained by performing feature extraction on the denoised first operating parameter based on a differential method.

[0089] Step 203, inputting the shape feature into the target fault diagnosis model to determine target fault coding information of the target nuclear power plant.

[0090] In a possible implementation manner, after the shape feature is input into the target fault diagnosis model, the shape feature can be processed by a Transformer model in the target fault diagnosis model to obtain key features of multi-dimensional time sequence features, and then the target fault coding information can be calculated through the target fully connected neural network.

[0091] In one embodiment, as Figure 3As shown, the similarity analysis on the first operation parameter based on the third operation parameter to obtain the second similarity of the first operation parameter includes the following steps:

[0092] Step 301, determining the fault type of the target nuclear power plant according to the target fault coding information;

[0093] Step 302, determining the third operation parameter according to the fault type of the target nuclear power plant.

[0094] In a possible implementation, assuming that the fault type of the target nuclear power plant determined based on the target fault coding information is type A, assuming that there are three fault reasons that can cause the A-type fault, which are A1, A2 and A3, respectively, obtaining the A1 operation parameter of the target nuclear power plant in the A1 fault condition, the A2 operation parameter of the target nuclear power plant in the A2 fault condition, and the A3 operation parameter of the target nuclear power plant in the A3 fault condition based on the fault parameter database, and determining the A1 operation parameter, the A2 operation parameter and the A3 operation parameter as the third operation parameter.

[0095] Step 303, determining the similarity of the first operation parameter and the third operation parameter based on the pattern distance method, and determining the similarity as the second similarity of the first operation parameter.

[0096] In a possible implementation, the A1 similarity, the A2 similarity and the A3 similarity of the first operation parameter and the A1 operation parameter, the A2 operation parameter and the A3 operation parameter can be respectively determined based on the pattern distance method, and the A1 similarity, the A2 similarity and the A3 similarity are determined as the second similarity of the first operation parameter.

[0097] In an optional embodiment of the present application, as described above, the operation parameter with the second similarity greater than the second similarity threshold value can be determined as the second target operation parameter, and the second target operation parameter is displayed.

[0098] In a possible implementation, assuming that only the A1 similarity in the second similarity is greater than the second similarity threshold value, the A1 operation parameter is determined as the second target operation parameter, and the second target operation parameter and A1 are displayed, so that the technical personnel can determine the specific fault reason of the target nuclear power plant according to the displayed second target operation parameter and A1.

[0099] In another possible implementation manner, assuming that the A2 similarity and the A3 similarity in the second similarity are both greater than the second similarity threshold, the A2 operation parameter and the A3 operation parameter are determined as the second target operation parameter, and the second target operation parameter, the A2, and the A3 are displayed, so that the technical personnel can determine the specific failure cause of the target nuclear power plant according to the displayed second target operation parameter, the A2, and the A3.

[0100] In one embodiment, as shown in FIG. 1, the training method of the target fault diagnosis model comprises the following steps: Figure 4

[0101] Step 401, obtaining a training data set comprising a plurality of labeled normal state operation parameters and labeled fault state operation parameters.

[0102] The normal state operation parameter is an operation parameter obtained when the target nuclear power plant is in a normal operation state, and the fault state operation parameter is an operation parameter obtained when the target nuclear power plant is in a failure state.

[0103] In one possible implementation manner, the historical data of the nuclear power plant can be obtained based on a database to obtain the normal state operation parameter and the fault state operation parameter, and the normal state operation parameter and the fault state operation parameter are labeled to determine the training data set.

[0104] In another possible implementation manner, the normal state operation parameter can also be obtained by using a monitoring system of an advanced data control room of the nuclear power plant, and the reactor nuclear power, the reactor thermal power, the feedwater temperature, the feedwater pressure, the feedwater flow, the steam temperature, the steam pressure, the steam flow, the positive rate of change of nuclear power, and the negative rate of change of nuclear power in the case of uncontrolled lifting of the control rod, the reactor nuclear power, the reactor thermal power, the feedwater temperature, the feedwater pressure, the feedwater flow, the steam temperature, the steam pressure, the steam flow, the positive rate of change of nuclear power, and the negative rate of change of nuclear power in the case of loss of all plant power, the reactor nuclear power, the reactor thermal power, the feedwater temperature, the feedwater pressure, the feedwater flow, the steam temperature, the steam pressure, the steam flow, the positive rate of change of nuclear power, and the negative rate of change of nuclear power in the case of failure of the feedwater regulating valve to close, the reactor nuclear power, the reactor thermal power, the feedwater temperature, the feedwater pressure, the feedwater flow, the steam temperature, the steam pressure, the steam flow, the positive rate of change of nuclear power, and the negative rate of change of nuclear power in the case of abnormal leakage of the primary coolant, and the reactor nuclear power, the reactor thermal power, the feedwater temperature, the feedwater pressure, the feedwater flow, the steam temperature, the steam pressure, the steam flow, the positive rate of change of nuclear power, and the negative rate of change of nuclear power in the case of leakage flow of the secondary heat transfer tube rupture are obtained as the fault state operation parameter, and the normal state operation parameter and the fault state operation parameter are labeled to determine the training data set. ​

[0105] Step 402: Determine the target fault diagnosis model based on the training dataset and the initial fault diagnosis model.

[0106] Optionally, the initial fault diagnosis model consists of a cascaded initial Transformer model and multiple initial fully connected neural networks.

[0107] In one possible implementation, the running parameters in the training dataset are processed, that is, the running parameters in the training dataset are preprocessed, and the shape features of each running parameter in the training dataset are determined. Based on the shape features and the initial fault diagnosis model, the target fault diagnosis model is determined.

[0108] In one embodiment, such as Figure 5 As shown, the determination of the target fault diagnosis model based on the training dataset and the initial fault diagnosis model includes the following steps:

[0109] Step 501: Determine a small sample dataset based on the training dataset, and perform small sample learning training based on the small sample dataset and the initial fault diagnosis model to obtain the first fault diagnosis model.

[0110] In one possible implementation, assume there are K types of faults, and the training dataset D = {(x1, y1), ..., (x... N y N There are N sets of operating parameters, each set of operating parameters x i The fault type label is y i , and y i For each fault type, a small number of corresponding operating parameters are taken as a small sample training set and randomly divided into a support set δ and a query set Q according to a certain ratio. Then, the small sample dataset is processed to obtain the shape features of each operating parameter in the small sample dataset. Based on the shape features, the initial fault diagnosis model is trained by small sample learning to obtain the first fault diagnosis model.

[0111] Step 502: Update and optimize the first fault diagnosis model based on the cross-entropy loss function algorithm and momentum optimization algorithm to obtain the target fault diagnosis model.

[0112] In one possible implementation, the running parameters (x) for each set in the target training set are obtained. i y i Fault coding information f θ (x i), the fault coding information corresponding to the operating parameters of the same fault type is processed, which can be taking an average value, and the data processing result is used to represent the type C of the fault k , wherein k represents the fault type, and the first fault diagnosis model is updated and optimized based on a cross-entropy loss function algorithm and a momentum optimization algorithm. Specifically, the distance d(f θ (x i ) between the feature coding f k (x θ ) and the fault type c i is calculated, and a loss function J(θ) is calculated k , wherein k' refers to all fault types, and the parameters θ in the first fault diagnosis model are updated and optimized based on the momentum optimization algorithm to minimize the value of the loss function, so as to obtain a target fault diagnosis model.

[0113] In an optional embodiment of the present application, the fault type of the target nuclear power plant can be determined according to the target fault coding information, as described above.

[0114] In a possible implementation, as described above, in the training process, the fault coding information f i (x i ) of each set of operating parameters (x θ , y i ) in the target training set is obtained, the fault coding information corresponding to the operating parameters of the same fault type is processed, which can be taking an average value, and a corresponding interval is set based on the average value. When the fault coding indicated by the target fault coding information falls within the interval, it can be determined that the target nuclear power plant is of the fault type corresponding to the interval.

[0115] In another possible implementation, as described above, in the training process, the fault coding information f i (x i ) of each set of operating parameters (x θ , y i ) in the target training set is obtained, the fault coding information corresponding to the operating parameters of the same fault type is processed, which can be taking an average value, and the value is determined as a fault standard value. Different fault types correspond to different fault standard values. The target fault coding is determined based on the target fault information, and the fault type corresponding to the fault standard value with the highest similarity to the target fault coding is determined as the fault type of the target nuclear power plant.

[0116] ​The method can train a fault diagnosis model with high accuracy using less sample data, realize small sample learning, reduce the cost of obtaining training data, improve the accuracy of the fault diagnosis model, and effectively improve the work efficiency of the fault diagnosis task.

[0117] In one embodiment, as shown in Figure 6 Another fault analysis method is provided, including the following steps:

[0118] Step 601, performing noise reduction processing on the first operating parameter based on an amplitude limiting value filtering algorithm and a Kalman filtering algorithm; performing feature extraction on the noise-reduced first operating parameter based on a piecewise linear representation method to obtain a shape feature of the first operating parameter; and inputting the shape feature into the target fault diagnosis model to determine target fault code information of the target nuclear power plant.

[0119] Step 602, in the presence of the target fault code information, determining the similarity between the first operating parameter and the second operating parameter based on a dynamic time warping method, and determining the similarity as a first similarity of the first operating parameter, the second operating parameter being an operating parameter of the target nuclear power plant in a normal state of the target nuclear power plant;

[0120] Step 603, determining the fault type of the target nuclear power plant according to the target fault code information; determining the third operating parameter according to the fault type of the target nuclear power plant; and determining the similarity between the first operating parameter and the third operating parameter based on a mode distance method, and determining the similarity as a second similarity of the first operating parameter.

[0121] Step 604, determining the first target operating parameter and the second target operating parameter according to the first similarity threshold, the second similarity threshold, the first similarity and the second similarity, and displaying the first target operating parameter and the second target operating parameter.

[0122] It should be understood that although the steps in the flowcharts involved in the embodiments described above are shown in sequence according to the arrows, the steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, the execution of the steps is not strictly limited in sequence, and the steps can be executed in other orders. Moreover, at least some of the steps in the flowcharts involved in the embodiments described above can include multiple steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution order of the steps or stages is not necessarily sequential, but can be alternately executed with other steps or steps or stages in other steps.

[0123] Based on the same inventive concept, the embodiments of the present application also provide a fault analysis device for implementing the above-mentioned fault analysis method. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme described in the above method, and therefore the specific limitations in one or more fault analysis device embodiments provided below can refer to the limitations of the fault analysis method described above, which will not be repeated here.

[0124] In an exemplary embodiment, as shown in Figure 7 a fault analysis device 700 is provided, comprising a determination module 701, a first execution module 702, a second execution module 703, and a third execution module 704, wherein:

[0125] The determination module 701 is configured to determine target fault code information of a target nuclear power plant according to a first operating parameter of the target nuclear power plant and a target fault diagnosis model, the target fault code information being used to indicate a fault type of the target nuclear power plant.

[0126] The first execution module 702 is configured to, in the case that the target fault code information exists, perform similarity analysis on the first operating parameter based on a second operating parameter to obtain a first similarity of the first operating parameter, the second operating parameter being an operating parameter of the target nuclear power plant in a normal state of the target nuclear power plant.

[0127] The second execution module 703 is configured to perform similarity analysis on the first operating parameter based on a third operating parameter to obtain a second similarity of the first operating parameter, the third operating parameter being an operating parameter with the same fault code information as the target fault code information.

[0128] The third execution module 704 is configured to determine a first target operating parameter and a second target operating parameter according to a first similarity threshold, a second similarity threshold, the first similarity, and the second similarity, and display the first target operating parameter and the second target operating parameter.

[0129] In one embodiment, the determining module 701 is specifically used to perform noise reduction processing on the first operating parameter based on the amplitude limiting filtering algorithm and the Kalman filtering algorithm; to extract features from the noise-reduced first operating parameter based on the piecewise linear representation method to obtain the shape features of the first operating parameter; and to input the shape features into the target fault diagnosis model to determine the target fault coding information of the target nuclear power plant.

[0130] In one embodiment, the first execution module 702 is specifically used to determine the similarity between the first operating parameter and the second operating parameter based on the dynamic time normalization method, and to determine the similarity as the first similarity of the first operating parameter.

[0131] In one embodiment, the second execution module 703 is specifically configured to determine the fault type of the target nuclear power plant based on the target fault coding information; determine the third operating parameter based on the fault type of the target nuclear power plant; determine the similarity between the first operating parameter and the third operating parameter based on the pattern distance method, and determine the similarity as the second similarity of the first operating parameter.

[0132] In one embodiment, such as Figure 8 As shown, another fault analysis device 800 is provided, which includes, in addition to the various modules included in the fault analysis device 700, a fourth execution module 705.

[0133] In one embodiment, the fourth execution module 705 is used to acquire a training dataset, which includes multiple labeled normal state operating parameters and labeled fault state operating parameters. The normal state operating parameters are operating parameters acquired when the target nuclear power plant is in normal operating condition, and the fault state operating parameters are operating parameters acquired when the target nuclear power plant is in fault state. The target fault diagnosis model is determined based on the training dataset and the initial fault diagnosis model.

[0134] In one embodiment, the fourth execution module 705 is specifically used to determine a small sample dataset based on the training dataset, and to perform small sample learning training based on the small sample dataset and the initial fault diagnosis model to obtain a first fault diagnosis model; and to update and optimize the first fault diagnosis model based on the cross-entropy loss function algorithm and the momentum optimization algorithm to obtain a target fault diagnosis model.

[0135] Each module in the aforementioned fault analysis device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0136] In an example embodiment, a computer device, which can be a server, is provided, and an internal structure diagram of the computer device can be as shown in Figure 9 The computer device includes a processor, a memory, an input / output interface (I / O), and a communication interface. The processor, the memory, and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for running the operating system and the computer program in the non-volatile storage medium. The database of the computer device is configured to store data. The input / output interface of the computer device is configured to exchange information between the processor and external devices. The communication interface of the computer device is configured to communicate with external terminals through a network connection. The computer program is executed by the processor to implement a fault analysis method.

[0137] Those skilled in the art can understand that Figure 9 The structure shown in the above is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the diagram, or combine certain components, or have a different arrangement of components.

[0138] In an example embodiment, a computer device is provided, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the following steps:

[0139] The target fault coding information is used to indicate a fault type of the target nuclear power plant. In a case where the target fault coding information exists, a similarity analysis is performed on the first operating parameter based on a second operating parameter to obtain a first similarity of the first operating parameter, the second operating parameter being an operating parameter of the target nuclear power plant in a normal state. A similarity analysis is performed on the first operating parameter based on a third operating parameter to obtain a second similarity of the first operating parameter, the third operating parameter being an operating parameter with the same fault coding information as the target fault coding information. First and second target operating parameters are determined according to a first similarity threshold, a second similarity threshold, the first similarity, and the second similarity, and the first and second target operating parameters are displayed.

[0140] In one of the embodiments, the processor, when executing the computer program, further implements the following steps: performing noise reduction processing on the first operation parameter based on a clipping value filtering algorithm and a Kalman filtering algorithm; performing feature extraction on the first operation parameter after the noise reduction processing based on a piecewise linear representation method to obtain a shape feature of the first operation parameter; and inputting the shape feature into the target fault diagnosis model to determine the target fault coding information of the target nuclear power plant.

[0141] In one of the embodiments, the processor, when executing the computer program, further implements the following steps: determining the similarity of the first operation parameter and the second operation parameter based on a dynamic time warping method, and determining the similarity as the first similarity of the first operation parameter.

[0142] In one of the embodiments, the processor, when executing the computer program, further implements the following steps: determining the fault type of the target nuclear power plant according to the target fault coding information; determining the third operation parameter according to the fault type of the target nuclear power plant; determining the similarity of the first operation parameter and the third operation parameter based on a pattern distance method, and determining the similarity as the second similarity of the first operation parameter.

[0143] In one of the embodiments, the processor, when executing the computer program, further implements the following steps: obtaining a training data set, the training data set including a plurality of labeled normal state operation parameters and labeled fault state operation parameters, the normal state operation parameters being operation parameters obtained when the target nuclear power plant is in a normal operation state, and the fault state operation parameters being operation parameters obtained when the target nuclear power plant is in a fault state; and determining the target fault diagnosis model according to the training data set and an initial fault diagnosis model.

[0144] In one of the embodiments, the processor, when executing the computer program, further implements the following steps: determining a small sample data set according to the training data set, and performing small sample learning training based on the small sample data set and the initial fault diagnosis model to obtain a first fault diagnosis model; and performing update optimization processing on the first fault diagnosis model based on a cross-entropy loss function algorithm and a momentum optimization algorithm to obtain a target fault diagnosis model.

[0145] In one embodiment, a computer readable storage medium is provided, having stored thereon a computer program which, when executed by a processor, implements the following steps:

[0146] The target fault coding information of the target nuclear power plant is determined according to the first operation parameter of the target nuclear power plant and a target fault diagnosis model, the target fault coding information is used to indicate a fault type of the target nuclear power plant; in the case that the target fault coding information exists, a similarity analysis is performed on the first operation parameter based on a second operation parameter to obtain a first similarity of the first operation parameter, the second operation parameter is an operation parameter of the target nuclear power plant in a normal state; a similarity analysis is performed on the first operation parameter based on a third operation parameter to obtain a second similarity of the first operation parameter, the third operation parameter is an operation parameter with the same fault coding information as the target fault coding information; first and second target operation parameters are determined according to a first similarity threshold, a second similarity threshold, the first similarity and the second similarity, and the first and second target operation parameters are displayed.

[0147] In one of the embodiments, the target fault coding information of the target nuclear power plant is determined according to the first operation parameter of the target nuclear power plant and a target fault diagnosis model, and the computer program, when executed by a processor, further implements the following steps: the first operation parameter is denoised based on a clipping value filtering algorithm and a Kalman filtering algorithm; the denoised first operation parameter is feature extracted based on a piecewise linear representation method to obtain a shape feature of the first operation parameter; and the shape feature is input into the target fault diagnosis model to determine the target fault coding information of the target nuclear power plant.

[0148] In one of the embodiments, the first similarity of the first operation parameter is obtained by performing a similarity analysis on the first operation parameter based on a second operation parameter, and the computer program, when executed by a processor, further implements the following steps: a similarity of the first operation parameter and the second operation parameter is determined based on a dynamic time warping method, and the similarity is determined as the first similarity of the first operation parameter.

[0149] In one of the embodiments, the second similarity of the first operation parameter is obtained by performing a similarity analysis on the first operation parameter based on a third operation parameter, and the computer program, when executed by a processor, further implements the following steps: a fault type of the target nuclear power plant is determined according to the target fault coding information; the third operation parameter is determined according to the fault type of the target nuclear power plant; a similarity of the first operation parameter and the third operation parameter is determined based on a pattern distance method, and the similarity is determined as the second similarity of the first operation parameter.

[0150] In one of the embodiments, the method for training the target fault diagnosis model, when the computer program is executed by the processor, further comprises the following steps: obtaining a training data set, the training data set comprising a plurality of labeled normal state operating parameters and labeled fault state operating parameters, the normal state operating parameters being operating parameters obtained when the target nuclear power plant is in a normal operating state, and the fault state operating parameters being operating parameters obtained when the target nuclear power plant is in a fault state; and determining the target fault diagnosis model according to the training data set and an initial fault diagnosis model.

[0151] In one of the embodiments, the method for determining the target fault diagnosis model according to the training data set and the initial fault diagnosis model, when the computer program is executed by the processor, further comprises the following steps: determining a small sample data set according to the training data set, and performing small sample learning training based on the small sample data set and the initial fault diagnosis model to obtain a first fault diagnosis model; and performing update optimization processing on the first fault diagnosis model based on a cross-entropy loss function algorithm and a momentum optimization algorithm to obtain the target fault diagnosis model.

[0152] In one of the embodiments, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the following steps:

[0153] determining target fault coding information of the target nuclear power plant according to the first operating parameters of the target nuclear power plant and the target fault diagnosis model, the target fault coding information being used to indicate a fault type of the target nuclear power plant; when the target fault coding information exists, performing similarity analysis on the first operating parameters based on second operating parameters to obtain a first similarity of the first operating parameters, the second operating parameters being operating parameters of the target nuclear power plant in a normal state; performing similarity analysis on the first operating parameters based on third operating parameters to obtain a second similarity of the first operating parameters, the third operating parameters being operating parameters with the same fault coding information as the target fault coding information; determining first target operating parameters and second target operating parameters according to a first similarity threshold, a second similarity threshold, the first similarity and the second similarity, and displaying the first target operating parameters and the second target operating parameters.

[0154] In one of the embodiments, the method for determining the target fault coding information of the target nuclear power plant according to the first operating parameters of the target nuclear power plant and the target fault diagnosis model, when the computer program is executed by the processor, further comprises the following steps: performing noise reduction processing on the first operating parameters based on a limiting value filtering algorithm and a Kalman filtering algorithm; performing feature extraction on the first operating parameters after the noise reduction processing based on a piecewise linear representation method to obtain shape features of the first operating parameters; and inputting the shape features into the target fault diagnosis model to determine the target fault coding information of the target nuclear power plant.

[0155] In one of the embodiments, the first operation parameter is analyzed for similarity based on the second operation parameter to obtain a first similarity of the first operation parameter, and the computer program, when executed by the processor, further implements the following steps: determining the similarity of the first operation parameter and the second operation parameter based on a dynamic time warping method, and determining the similarity as the first similarity of the first operation parameter.

[0156] In one of the embodiments, the first operation parameter is analyzed for similarity based on the third operation parameter to obtain a second similarity of the first operation parameter, and the computer program, when executed by the processor, further implements the following steps: determining a fault type of the target nuclear power plant according to the target fault coding information; determining the third operation parameter according to the fault type of the target nuclear power plant; determining the similarity of the first operation parameter and the third operation parameter based on a pattern distance method, and determining the similarity as the second similarity of the first operation parameter.

[0157] In one of the embodiments, the training method of the target fault diagnosis model, the computer program, when executed by the processor, further implements the following steps: obtaining a training data set, the training data set including a plurality of labeled normal state operation parameters and labeled fault state operation parameters, the normal state operation parameters being operation parameters obtained when the target nuclear power plant is in a normal operation state, and the fault state operation parameters being operation parameters obtained when the target nuclear power plant is in a fault state; and determining the target fault diagnosis model according to the training data set and an initial fault diagnosis model.

[0158] In one of the embodiments, the target fault diagnosis model is determined according to the training data set and the initial fault diagnosis model, and the computer program, when executed by the processor, further implements the following steps: determining a small sample data set according to the training data set, and performing small sample learning training based on the small sample data set and the initial fault diagnosis model to obtain a first fault diagnosis model; and performing update optimization processing on the first fault diagnosis model based on a cross-entropy loss function algorithm and a momentum optimization algorithm to obtain a target fault diagnosis model.

[0159] It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of the relevant data need to comply with relevant regulations.

[0160] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the processes of the above-mentioned embodiments of the methods can be included. Any reference to memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (Read-Only Memory, ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetoresistive random access memory (Magnetoresistive Random Access Memory, MRAM), ferroelectric memory (Ferroelectric Random Access Memory, FRAM), phase change memory (Phase Change Memory, PCM), graphene memory, etc. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory, etc. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM), etc. The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a block chain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., without being limited thereto.

[0161] Any combination of the technical features of the above embodiments can be made. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combination of the technical features does not exist contradictory, it should be considered as the scope of the present application.

[0162] The above embodiments only express several implementation manners of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the patent of the present application. It should be pointed out that for ordinary skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of protection of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A fault analysis method, characterized in that, The method includes: The target fault coding information of the target nuclear power plant is determined based on the first operating parameters of the target nuclear power plant and the target fault diagnosis model. The target fault coding information is used to indicate the fault type of the target nuclear power plant. The target fault diagnosis model consists of a cascaded target Transformer model and two target fully connected neural networks. In the presence of the target fault coding information, the similarity between the first operating parameter and the second operating parameter is determined based on the dynamic time-normalization method, and the similarity is determined as the first similarity of the first operating parameter, and the second operating parameter is the operating parameter of the target nuclear power plant under normal conditions; The fault type of the target nuclear power plant is determined based on the target fault coding information; a third operating parameter is determined based on the fault type; the similarity between the first operating parameter and the third operating parameter is determined based on the pattern distance method, and the similarity is determined as the second similarity of the first operating parameter; the third operating parameter is the operating parameter whose fault coding information is the same as the target fault coding information. The first target operating parameter and the second target operating parameter are determined based on the first similarity threshold, the second similarity threshold, the first similarity, and the second similarity, and are displayed. The first target operating parameter is the operating parameter whose first similarity is less than the first similarity threshold. The second target operating parameter is the operating parameter whose second similarity is greater than the second similarity threshold. The training method for the target fault diagnosis model includes: acquiring a training dataset, which includes multiple labeled normal state operating parameters and labeled fault state operating parameters. The normal state operating parameters are operating parameters acquired when the target nuclear power plant is in normal operating condition, and the fault state operating parameters are operating parameters acquired when the target nuclear power plant is in fault state. A small sample dataset is determined based on the training dataset; the small sample dataset is processed to obtain the shape features of each running parameter in the small sample dataset, and the initial fault diagnosis model is trained using small sample learning based on the shape features to obtain a first fault diagnosis model; the first fault diagnosis model is updated and optimized based on the cross-entropy loss function algorithm and the momentum optimization algorithm to obtain the target fault diagnosis model.

2. The method according to claim 1, characterized in that, The step of determining the target fault coding information of the target nuclear power plant based on the first operating parameters of the target nuclear power plant and the target fault diagnosis model includes: The first operating parameter is denoised based on the amplitude limiting filtering algorithm and the Kalman filtering algorithm; Based on the piecewise linear representation method, feature extraction is performed on the first operating parameter after the noise reduction process to obtain the shape feature of the first operating parameter; The shape features are input into the target fault diagnosis model to determine the target fault coding information of the target nuclear power plant.

3. The method according to claim 1, characterized in that, The first operating parameters include reactor nuclear power, reactor thermal power, feedwater temperature, feedwater pressure, feedwater flow rate, steam temperature, steam pressure, steam flow rate, positive rate of change of nuclear power, and negative rate of change of nuclear power.

4. The method according to claim 1, characterized in that, The first target operating parameters are used to determine the equipment in the target nuclear power plant that is in a faulty state.

5. The method according to claim 1, characterized in that, The second target operating parameters are used to determine the specific cause of failure in the target nuclear power plant.

6. The method according to claim 2, characterized in that, The noise reduction processing of the first operating parameter based on the amplitude limiting filtering algorithm and the Kalman filtering algorithm includes: The first operating parameters are denoised based on the amplitude limiting filtering algorithm to obtain the first operating parameters after denoising based on the amplitude limiting filtering algorithm. The first operating parameters after noise reduction processing based on the amplitude limiting filter algorithm are further denoised using the Kalman filter algorithm to obtain the first operating parameters after noise reduction processing.

7. A fault analysis device, characterized in that, The device includes: The determination module is used to determine the target fault coding information of the target nuclear power plant based on the first operating parameters of the target nuclear power plant and the target fault diagnosis model. The target fault coding information is used to indicate the fault type of the target nuclear power plant. The target fault diagnosis model consists of a cascaded target Transformer model and two target fully connected neural networks. The first execution module is used to determine the similarity between the first operating parameter and the second operating parameter based on the dynamic time-normalization method when the target fault coding information exists, and to determine the similarity as the first similarity of the first operating parameter, wherein the second operating parameter is the operating parameter of the target nuclear power plant under normal conditions; The second execution module is used to determine the fault type of the target nuclear power plant based on the target fault coding information; determine the third operating parameter based on the fault type; determine the similarity between the first operating parameter and the third operating parameter based on the pattern distance method, and determine the similarity as the second similarity of the first operating parameter, wherein the third operating parameter is an operating parameter whose fault coding information is the same as the target fault coding information; The third execution module is used to determine a first target running parameter and a second target running parameter from the first running parameters based on a first similarity threshold, a second similarity threshold, the first similarity, and the second similarity, and to display the first target running parameter and the second target running parameter; the first target running parameter is the running parameter among the first running parameters whose first similarity is less than the first similarity threshold; the second target running parameter is the running parameter among the first running parameters whose second similarity is greater than the second similarity threshold; The fourth execution module is used to acquire a training dataset, which includes multiple labeled normal operating parameters and labeled fault operating parameters. The normal operating parameters are obtained when the target nuclear power plant is in normal operating condition, and the fault operating parameters are obtained when the target nuclear power plant is in fault condition. The module then determines a small sample dataset based on the training dataset; processes the small sample dataset to obtain the shape features of each operating parameter in the small sample dataset; and trains an initial fault diagnosis model using small sample learning based on the shape features to obtain a first fault diagnosis model. Finally, it updates and optimizes the first fault diagnosis model using a cross-entropy loss function algorithm and a momentum optimization algorithm to obtain the target fault diagnosis model.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images