Nuclear reactor fault classification method, device, computer equipment and storage medium

Abstract

This application relates to a method, device, computer equipment, storage medium and computer program product for classifying nuclear reactor faults. The method includes: obtaining the state best estimate data of the nuclear reactor at different sampling moments, where the state best estimate data includes data in multiple dimensions; generating state best estimate sequences in multiple dimensions based on a preset period according to the state best estimate data; constructing a state estimation matrix according to the state best estimate sequences in multiple dimensions; inputting the state estimation matrix into a preset fault classification model to obtain a state fault classification result and the confidence level corresponding to the state fault classification result; and inputting the state fault classification result and the confidence level corresponding to the state fault classification result into a preset self-learning model to obtain the fault classification result of the nuclear reactor and the confidence level corresponding to the fault classification result. By using this method, a high-precision nuclear reactor fault classification result can be obtained.

Description

Technical Field

[0001] The present application relates to the technical field of nuclear reactor fault diagnosis, and in particular to a nuclear reactor fault classification method, device, computer equipment, storage medium and computer program product. Background Art

[0002] Fault diagnosis, as a multidisciplinary technology involving measuring instruments and control, industrial big data, artificial intelligence, reliability, quality engineering, etc., has been widely used in aerospace, rail transportation, energy equipment, intelligent factories, etc. Specifically in the field of nuclear reactor systems, nuclear reactor systems are complex dynamic systems with strict safety restrictions, making it difficult and crucial to diagnose all types of faults in the early stage. Its key safety goal is to correctly and accurately identify the state when any abnormality / fault occurs, assist operators in making correct decisions, and thus improve the safety level of these reactors. Considering the catastrophic consequences that nuclear accidents can cause, a diagnostic system that classifies deviations from the normal operation of nuclear power plants can help operators focus on the most important monitoring, decision-making, and control, and reduce the operator's decision-making error rate and workload. Therefore, it is necessary to study how to improve the diagnostic accuracy of nuclear reactor fault types.

[0003] Early fault diagnosis systems based on expert system technology had obvious limitations. Traditional nuclear reactor accident diagnosis systems use expert system technology, that is, through "yes, no" logical judgments on a series of processes of nuclear reactor system fault diagnosis, they provide support for tasks such as status monitoring, fault diagnosis, and operation and maintenance decisions. Because of the use of rule-based judgment methods, certain symptoms are judged, but there is no deep understanding of the system mechanism and it is impossible to handle situations outside the rules. Although some traditional fuzzy logic rules can "guess" about new situations outside the rules, there is still great uncertainty. Coupled with the lack of expert knowledge base, these early systems have difficulty in overcoming their effectiveness problems. In the past two decades, new fault diagnosis methods have been widely studied and applied in nuclear energy systems. Existing fault diagnosis methods are extremely sensitive to measurement noise, and when they encounter information that does not belong to their training or knowledge reserve type, they overfit or out-of-range reasoning occurs, and they do not have the ability to respond to "I don't know". Deep learning has low robustness in the presence of uncertain disturbances or poor data quality, which affects the accuracy of fault diagnosis. Moreover, existing learning-based methods, especially machine learning methods based on shallow representation, cannot fully meet the accuracy requirements of fault classification diagnosis.

[0004] It can be seen that the technical solution of the above-mentioned prior art has the problem of low fault classification accuracy. Summary of the invention

[0005] Based on this, it is necessary to provide a more accurate nuclear reactor fault classification method, device, computer equipment, computer readable storage medium and computer program product to address the above technical problems.

[0006] In a first aspect, the present application provides a method for classifying nuclear reactor faults. The method comprises:

[0007] Obtaining optimal state estimation data of the nuclear reactor at different sampling times, wherein the optimal state estimation data includes data in multiple dimensions;

[0008] Generate a state best estimation sequence of multiple dimensions based on a preset period according to the state best estimation data;

[0009] According to the state best estimation sequence of multiple dimensions, a state estimation matrix is ​​formed;

[0010] Input the state estimation matrix into a preset fault classification model to obtain a state fault classification result and a confidence level corresponding to the state fault classification result;

[0011] The state fault classification result and the confidence level corresponding to the state fault classification result are input into a preset self-learning model to obtain the fault classification result of the nuclear reactor and the confidence level corresponding to the fault classification result.

[0012] In one embodiment, obtaining optimal estimation data of the state of the nuclear reactor at different sampling times includes:

[0013] Obtaining status data of nuclear reactors at different sampling times;

[0014] The state data at different sampling times are fused through the state optimal estimation fusion algorithm to obtain the state optimal estimation data of the nuclear reactor at different sampling times.

[0015] In one embodiment, the status data includes:

[0016] High-precision simulation status data, real-time simulation status data, real-time measurement status data, and periodic measurement status data.

[0017] In one embodiment, the state data at different sampling times are fused by using the state optimal estimation fusion algorithm to obtain the state optimal estimation data of the nuclear reactor at different sampling times, including:

[0018] Obtaining the best estimated data of the state of the nuclear reactor at the previous moment at different sampling moments and the confidence level corresponding to the best estimated data of the state at the previous moment;

[0019] According to the best estimated data of the state at the previous moment and the confidence level corresponding to the best estimated data of the state at the previous moment, the state prediction data of the nuclear reactor at different sampling moments are obtained;

[0020] Obtaining state measurement data of the nuclear reactor at different sampling times;

[0021] The state prediction data and the state measurement data are fused through the state optimal estimation fusion algorithm to obtain the state optimal estimation data of the nuclear reactor at different sampling times.

[0022] In one embodiment, obtaining the state prediction data of the nuclear reactor at different sampling times according to the state best estimation data at the previous moment and the confidence level corresponding to the state best estimation data at the previous moment includes:

[0023] According to the best estimated state data at the previous moment and the confidence level of the best estimated state data, the state data of particles in the nuclear reactor at different sampling moments at the previous moment are obtained;

[0024] Inputting the state data of the particles at the previous moment at different sampling moments into a preset simulation prediction model, and outputting the state prediction data of the particles at different sampling moments;

[0025] According to the state prediction data of particles at different sampling moments, the state prediction data of the nuclear reactor at different sampling moments are obtained.

[0026] In one embodiment, the state prediction data and the state measurement data are fused by a state optimal estimation fusion algorithm to obtain the state optimal estimation data of the nuclear reactor at different sampling times, including:

[0027] Obtain the confidence level corresponding to the state prediction data of the nuclear reactor at different sampling times;

[0028] Acquire state data of particles in the nuclear reactor at different sampling times according to the state prediction data and the confidence level corresponding to the state prediction data;

[0029] According to the state data, estimated state measurement data of the nuclear reactor at different sampling times and confidence levels corresponding to the estimated state measurement data are obtained;

[0030] Obtain actual state measurement data of the nuclear reactor at different sampling times;

[0031] According to the actual state measurement data and the estimated state measurement data, the residual coefficients of the state measurement data of the nuclear reactor at different sampling times are obtained;

[0032] According to the confidence level corresponding to the estimated state measurement data, a gain coefficient of the state measurement data is obtained;

[0033] According to the gain coefficient and the residual coefficient, the state prediction data and the state measurement data are fused to obtain the optimal state estimation data of the nuclear reactor at different sampling times.

[0034] In a second aspect, the present application also provides a nuclear reactor fault classification device. The device comprises:

[0035] A data acquisition module is used to acquire the best estimated state data of the nuclear reactor at different sampling times, and the best estimated state data includes data of multiple dimensions;

[0036] A data processing module generates a state best estimation sequence of multiple dimensions based on a preset period according to the state best estimation data; and forms a state estimation matrix according to the state best estimation sequence of multiple dimensions;

[0037] The fault classification module is used to input the state estimation matrix into a preset fault classification model to obtain the state fault classification result and the confidence level corresponding to the state fault classification result; input the state fault classification result and the confidence level corresponding to the state fault classification result into a preset self-learning model to obtain the fault classification result of the nuclear reactor and the confidence level corresponding to the fault classification result.

[0038] In a third aspect, the present application further provides a computer device. The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:

[0039] Obtaining optimal state estimation data of the nuclear reactor at different sampling times, wherein the optimal state estimation data includes data in multiple dimensions;

[0040] Generate a state best estimation sequence of multiple dimensions based on a preset period according to the state best estimation data;

[0041] According to the state best estimation sequence of multiple dimensions, a state estimation matrix is ​​formed;

[0042] Input the state estimation matrix into a preset fault classification model to obtain a state fault classification result and a confidence level corresponding to the state fault classification result;

[0043] The state fault classification result and the confidence level corresponding to the state fault classification result are input into a preset self-learning model to obtain the fault classification result of the nuclear reactor and the confidence level corresponding to the fault classification result.

[0044] In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0045] Obtaining optimal state estimation data of the nuclear reactor at different sampling times, wherein the optimal state estimation data includes data in multiple dimensions;

[0046] Generate a state best estimation sequence of multiple dimensions based on a preset period according to the state best estimation data;

[0047] According to the state best estimation sequence of multiple dimensions, a state estimation matrix is ​​formed;

[0048] Input the state estimation matrix into a preset fault classification model to obtain a state fault classification result and a confidence level corresponding to the state fault classification result;

[0049] The state fault classification result and the confidence level corresponding to the state fault classification result are input into a preset self-learning model to obtain the fault classification result of the nuclear reactor and the confidence level corresponding to the fault classification result.

[0050] In a fifth aspect, the present application further provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0051] Obtaining optimal state estimation data of the nuclear reactor at different sampling times, wherein the optimal state estimation data includes data in multiple dimensions;

[0052] Generate a state best estimation sequence of multiple dimensions based on a preset period according to the state best estimation data;

[0053] According to the state best estimation sequence of multiple dimensions, a state estimation matrix is ​​formed;

[0054] Input the state estimation matrix into a preset fault classification model to obtain a state fault classification result and a confidence level corresponding to the state fault classification result;

[0055] The state fault classification result and the confidence level corresponding to the state fault classification result are input into a preset self-learning model to obtain the fault classification result of the nuclear reactor and the confidence level corresponding to the fault classification result.

[0056] The above-mentioned nuclear reactor fault classification method, device, computer equipment, storage medium and computer program product, on the one hand, can obtain accurate fault classification results by obtaining the best estimation data of the state of the nuclear reactor at different sampling times and integrating the data at different sampling times, thereby greatly improving the accuracy of fault classification diagnosis; on the other hand, the state estimation matrix is ​​first input into a preset fault classification model to obtain the state fault classification result and the confidence corresponding to the fault classification result, and then the obtained state fault classification result and the confidence corresponding to the state fault classification result are input into a preset self-learning model, and different fault classification results are aggregated, which can ensure the accuracy of the final output nuclear reactor fault classification result and the confidence corresponding to the fault classification result, thereby obtaining a high-precision nuclear reactor fault classification result. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 A diagram of an application environment of a nuclear reactor fault classification method in one embodiment;

[0058] Figure 2 is a schematic flow chart of a method for classifying nuclear reactor faults in one embodiment;

[0059] Figure 3 A schematic diagram of a process of fusing state data to obtain optimal state estimation data in one embodiment;

[0060] Figure 4 A schematic diagram of nuclear reactor status data fusion in one embodiment;

[0061] Figure 5 This is a schematic diagram of the process of sub-step S380 in another embodiment;

[0062] Figure 6 A structural block diagram of a nuclear reactor fault classification method device in one embodiment;

[0063] Figure 7 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0064] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0065] The nuclear reactor fault classification method provided in the embodiment of the present application can be applied to Figure 1In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. The server 104 obtains the state best estimation data of the terminal 102 at different sampling times, generates the state best estimation sequence of multiple dimensions based on the preset period according to the obtained state best estimation data, and forms a state estimation matrix, inputs the state best estimation matrix into the preset fault classification model, obtains the state fault classification result and the corresponding confidence, inputs the obtained state fault classification result and its corresponding confidence into the preset self-learning model, and obtains the fault classification result of the nuclear reactor and its confidence. Among them, the terminal 102 can be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, Internet of Things devices and portable wearable devices, and the Internet of Things devices can be smart speakers, smart TVs, smart air conditioners, smart car-mounted devices, etc. Portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The server 104 can be implemented with an independent server or a server cluster consisting of multiple servers.

[0066] In one embodiment, Figure 2 As shown, a nuclear reactor fault classification method is provided, which is applied to Figure 1 The server in the example is used to illustrate the following steps:

[0067] S100, obtaining optimal state estimation data of a nuclear reactor at different sampling times, wherein the optimal state estimation data includes data of multiple dimensions.

[0068] Among them, the state best estimate data is the state estimate data that is closest to the real state inside the nuclear reactor. Specifically, the nuclear reactor includes many particles, each particle presents different states at different sampling times, and the state of all particles in the reactor can be used to characterize the state of the entire reactor at the sampling time. According to the state data of all particles in the nuclear reactor at different sampling times, fusion processing is performed to obtain the state best estimate data of the nuclear reactor at different sampling times, and there is a state best estimate data at each sampling time. At the same time, it should be noted that the state best estimate data of the nuclear reactor and the state data of the particles both include parameter data of multiple dimensions, such as: reactivity coefficient, deviation from nucleate boiling, maximum linear power density, maximum fuel rod and cladding temperature and other parameter data that are very relevant to the safe operation of the nuclear reactor. The above parameter data of multiple dimensions together constitute the state best estimate data or the state data of a particle.

[0069] S200, generating a state best estimation sequence of multiple dimensions based on a preset period according to the state best estimation data.

[0070] Among them, the state best estimation sequence is a sequence into which the state best estimation data at different sampling moments are divided according to a preset period. Specifically, according to the preset period, the state best data at different sampling moments are combined in the order of each moment, and a state best estimation sequence of multiple dimensions is obtained accordingly. For example, the state best estimation data of a nuclear reactor at 1s, 2s, 3s, 4s, and 5s is obtained, and each 3s is preset as a period, then three multi-dimensional state best estimation sequence sequences of {3s, 4s, 5s}, {2s, 3s, 4s}, and {1s, 2s, 3s} can be obtained.

[0071] S300, constructing a state estimation matrix according to the state best estimation sequences of multiple dimensions.

[0072] Among them, the state estimation matrix refers to the collection of data of multiple dimensions in the state best estimation data at each sampling moment in the same state best estimation sequence. Specifically, the diagnosis of accident / fault type based solely on the current state of the nuclear reactor is the mainstream technology for many fault diagnoses at present. However, if there is only data at the current moment and the historical change trend of the data is ignored, some extreme evolution accidents of the reactor operating conditions cannot be effectively diagnosed, and the opportunity for effective intervention in the reactor operation is missed. Therefore, in order to ensure the accuracy of the fault classification results obtained, it is very necessary to effectively fuse the data at the historical sampling moments for diagnosis. The importance of data at different historical sampling moments relative to the data at the current sampling moment has a decreasing effect, and the data at different historical sampling moments are formed into a matrix. The typical nuclear reactor state with an update frequency of 1 second can be preset to a period of 10 seconds to a few minutes. According to the comprehensive trade-offs such as the trend of changes in the nuclear reactor operating conditions and the matrix processing speed, it should be noted that the data in the state matrix should also be normalized, that is:

[0073]

[0074] Where P is the estimated value of the state variable at the sampling moment, and Pmin and Pmax represent the maximum and minimum values ​​of the state physical quantity under all accident conditions. Take the 5th second as the current sampling moment for example: three multi-dimensional state best estimation sequences {3s, 4s, 5s}, {2s, 3s, 4s}, and {1s, 2s, 3s} can be obtained. The state best estimation sequence {3s, 4s, 5s} is selected for explanation. The nuclear reactor has a state best estimation data at each moment in the state best estimation sequence. At each moment, 100 particles are extracted from the nuclear reactor as samples to obtain the state data of 100 particles at each moment. In the state best estimation sequence {3s, 4s, 5s}, 300 state data can be obtained accordingly. These 300 state data together constitute a state estimation matrix 1. Similarly, the other two sequences {2s, 3s, 4s} and {1s, 2s, 3s} can also obtain 300 state data respectively, forming the state estimation matrix 2 and the state optimal estimation matrix 3.

[0075] S400, inputting the state estimation matrix into a preset fault classification model to obtain a state fault classification result and a confidence level corresponding to the state fault classification result.

[0076] Among them, the preset fault classification model is a model of a neural network with a fault classification function pre-set; confidence refers to the probability that the value of the population parameter falls within a certain area of ​​the sample statistical value when sampling to estimate the population parameter. Specifically, the state of the nuclear reactor changes in real time, and the state monitoring data also changes continuously. Therefore, in order to ensure the accuracy of the state fault classification result, all state data obtained at different sampling times need to be used as the input of the preset fault classification model. Through deep learning in the deep residual contraction network in the preset fault classification model, the state fault classification result and the confidence corresponding to the state fault classification result can be effectively identified according to the multi-dimensional state data in the state estimation matrix, thereby realizing preliminary fault classification diagnosis. Similarly, the 5th second is used as the current moment for explanation. The state estimation matrix 1, state estimation matrix 2, and state estimation matrix 3 obtained at the 5th second are sequentially input into the deep residual contraction network in the preset fault classification model for deep learning, and the fault classification result 1, fault classification result 2, fault classification result 3 and the corresponding confidence of each fault classification result are output respectively. For further explanation, the deep residual shrinkage network in the preset fault classification model is described as follows: The main advantage of the deep residual shrinkage network over the ordinary deep learning method is that it is more suitable for feature extraction of noisy data. Because the deep residual shrinkage network uses soft thresholding as a nonlinear layer in its structure, it is the core step of signal denoising, which is equivalent to integrating denoising into the deep neural network and making it a trainable step.

[0077] S500, inputting the state fault classification result and the confidence level corresponding to the state fault classification result into a preset self-learning model to obtain the fault classification result of the nuclear reactor and the confidence level corresponding to the fault classification result.

[0078] Among them, the preset self-learning model refers to a model with self-learning function including multiple types of neural networks. Specifically, different state best estimation matrices come from different state best estimation sequences, and the state fault classification results obtained according to different state best estimation matrices may be inconsistent with each other. This is due to the combined superposition of measurement uncertainty, prediction result uncertainty, and uncertainty of deep neural network diagnosis model. In order to consider the differences between different measurement results, a self-learning model is constructed based on the idea of ​​ensemble learning stacking. This self-learning model is essentially a secondary learning model, which can be a machine learning model with strong random effects such as random forest, or a learning model composed of a forward artificial neural network. In the self-learning model, different types of nuclear reactor fault classification diagnosis models are placed in a framework based on an integrated learning method to achieve complementary advantages between different models, and the previously obtained state fault classification results and the confidence corresponding to the state fault classification results are input into a pre-built self-learning model. The stacking technology is used to take the previously obtained multiple state fault classification results and the confidence corresponding to the state fault classification results as the basic diagnosis results. Under large sample space variation or large uncertainty, the multiple state fault classification results and the confidence corresponding to the state fault classification results are stacked and combined, and the obtained nuclear reactor fault classification results and the confidence corresponding to the fault classification results are used as the final diagnosis result. For example: the obtained fault classification result 1, fault classification result 2, and fault classification result 3 are input into a preset self-learning model, which may include a deep residual network 1, a deep residual network 2 or more other deep neural networks, and a final fault classification result of a nuclear reactor at the current moment and the confidence of the fault classification result are output. In this process, multiple fault classification diagnosis results are stacked, so that the accuracy of the final diagnosis result is higher.

[0079] The above-mentioned nuclear reactor fault classification method, on the one hand, can obtain accurate fault classification results by obtaining the best estimation data of the state of the nuclear reactor at different sampling times and integrating the data at different sampling times, thereby greatly improving the accuracy of fault classification diagnosis; on the other hand, the state estimation matrix is ​​first input into a preset fault classification model to obtain the state fault classification results and the confidence corresponding to the fault classification results, and then the obtained state fault classification results and the confidence corresponding to the state fault classification results are input into a preset self-learning model, and different fault classification results are aggregated, which can ensure the accuracy of the final output nuclear reactor fault classification results and the confidence corresponding to the fault classification results, thereby obtaining a high-precision nuclear reactor fault classification result.

[0080] In one embodiment, obtaining optimal estimation data of the state of the nuclear reactor at different sampling times includes:

[0081] Step 1, obtaining the status data of the nuclear reactor at different sampling times;

[0082] Step 2: The state data at different sampling times are fused through the state optimal estimation fusion algorithm to obtain the state optimal estimation data of the nuclear reactor at different sampling times.

[0083] Among them, the state optimal estimation fusion algorithm is an algorithm that converts the optimal state estimation of multiple measurement data into the estimated value of the posterior probability of the state value that meets the measurement requirements based on the posterior estimation theory. Specifically, the state data of the nuclear reactor at different angles and different update cycles at different sampling times are obtained, which are mainly divided into two categories: simulation state data and measurement state data. The simulation state data and the measurement state data are fused through the state optimal estimation fusion algorithm to obtain the state optimal estimation data.

[0084] In this embodiment, the state data of the nuclear reactor at different sampling times are integrated through the state optimal estimation algorithm to obtain the state optimal estimation data, which is data extracted from state data at different angles and different update cycles. Compared with a single state data, it can more accurately estimate the state of the nuclear reactor system at each sampling time, and to a certain extent ensure that accurate fault classification results can be obtained subsequently, thereby improving the efficiency and accuracy of fault classification diagnosis.

[0085] In one embodiment, the status data includes:

[0086] High-precision simulation status data, real-time simulation status data, real-time measurement status data, and periodic measurement status data.

[0087] Among them, high-precision simulation state data is data generated by professional nuclear energy design software. These design software are based on the physical equations of the first principles and have been fully tested, verified and confirmed. These software give high-precision calculation results of nuclear energy systems under various conditions to ensure the safety of the designed nuclear reactor; real-time simulation state data is data obtained by real-time simulation models, and real-time simulation models are refined models of high-precision simulations that are reduced in order, simplified, and equivalent to achieve real-time simulation of reactors within acceptable engineering accuracy, such as coarser spatial grid division, and conversion of three-dimensional models to point pile models. Real-time measurement state data consists of thousands of measured hardware signals in the nuclear energy system and is refreshed at a certain frequency (such as 0.1s). The comparison and analysis of real-time measurement data and set values ​​(or curves) will serve as early warning, control and protection signals to achieve specific component or component level operations. Periodic measurement state data is higher-precision experimental measurement state data obtained periodically under given experimental conditions. Specifically, obtaining the above four different angles of state data in different ways at each sampling time can characterize the state of the nuclear reactor system to a certain extent. By comparing the above four types of data, it is possible to determine whether a fault has occurred in the nuclear reactor system, as well as the location and cause of the fault, because these four parameters include many hidden variables that cannot be directly measured but are very relevant to safe operation (such as reactivity coefficient, deviation from nucleate boiling, maximum linear power density, maximum fuel rod and cladding temperature, etc.).

[0088] In this embodiment, the status data of the nuclear reactor includes high-precision simulation status data, real-time simulation status data, real-time measurement status data and periodic measurement status data. The accuracy of the optimal status estimation data obtained after fusing the status data of the above four different angles and different update cycles is relatively high, which can further ensure the accuracy of the final nuclear reactor fault classification result.

[0089] In one embodiment, Figure 3 As shown, through the state optimal estimation fusion algorithm, the state data at different sampling times are fused to obtain the state optimal estimation data of the nuclear reactor at different sampling times, including:

[0090] S320, obtaining the best estimated data of the state of the nuclear reactor at the previous moment at different sampling moments and the confidence level corresponding to the best estimated data of the state at the previous moment;

[0091] S340, obtaining state prediction data of the nuclear reactor at different sampling times according to the best estimated state data at the previous moment and the confidence level corresponding to the best estimated state data at the previous moment;

[0092] S360, obtaining state measurement data of the nuclear reactor at different sampling times;

[0093] S380, the state prediction data and the state measurement data are fused through the state optimal estimation fusion algorithm to obtain the state optimal estimation data of the nuclear reactor at different sampling times.

[0094] Specifically, Figure 4 As shown, starting from the best estimate of the state of the nuclear reactor at the moment before each sampling moment T0, based on high-precision simulation or real-time simulation software, the predicted data and confidence (prior probability) of the nuclear reactor state at each sampling moment T1 can be obtained. Then, after obtaining the measurement data of the nuclear reactor state at moment T1, it is necessary to carry out a fusion calculation of the two based on the uncertainty of the measurement and the prior confidence of the predicted data to obtain the best estimated state data and its uncertainty (posterior probability) at each sampling moment T1. Then, starting from moment T1, estimate the state posteriori at moment T2 until the tracking of the entire reactor operation is completed.

[0095] In this embodiment, the state prediction data at the sampling moment is obtained based on the state optimal data at the moment before the sampling moment, and the obtained state prediction data is integrated with the acquired state measurement data, which can greatly improve the efficiency of data processing, obtain more accurate state optimal estimation data, and ensure the accuracy of the final nuclear reaction final fault classification result.

[0096] In one embodiment, according to the state best estimation data at the previous moment and the confidence level corresponding to the state best estimation data at the previous moment, the state prediction data of the nuclear reactor at different sampling moments is obtained, including:

[0097] Step 1, obtaining the state data of particles in the nuclear reactor at different sampling times at the previous moment according to the state best estimation data at the previous moment and the confidence of the state best estimation data;

[0098] Step 2, inputting the state data of the particle at the previous moment at different sampling moments into a preset simulation prediction model, and outputting the state prediction data of the particle at different sampling moments;

[0099] Step 3, obtaining the state prediction data of the nuclear reactor at different sampling times according to the state prediction data of the particles at different sampling times.

[0100] The state prediction data is the state data of the current moment predicted based on the previous moment. Specifically, the best estimate of the state of the nuclear reactor at the time t before the given sampling moment is s t and its confidence (covariance matrix) P t , the average prediction error e of the known simulation prediction model (including high-precision simulation model or real-time simulation model) m Under the premise of and confidence (covariance matrix) Q, N particles are extracted from the nuclear reactor to obtain the state data of each particle at the previous moment of each sampling moment Where φ is a sampling function of a multivariate Gaussian distribution, i represents the i-th particle, and N particles representing the prediction error of the nuclear reactor state are constructed. Where φ also represents the sampling function of the multivariate Gaussian distribution. The state data of each particle at the moment before the sampling moment is sequentially input into the preset simulation prediction model, and the state prediction data of each particle at each sampling moment is output. Used to predict the state of each particle at each sampling time, where F represents the accuracy of the high-precision or low-precision simulation model. Predict data based on the state of the extracted particles at each sampling time Obtain the state prediction data of the entire nuclear reactor at each sampling time It is used to predict the state of the nuclear reactor at the sampling time, which can be calculated using the following formula:

[0101]

[0102] In this embodiment, the state estimation data of the particles in the nuclear reactor at the moment before the sampling moment is input into the preset simulation prediction model, so that the state prediction data of the particles in the nuclear reactor at the sampling moment can be quickly obtained, and the state prediction data of each particle in the nuclear reactor can be aggregated to obtain accurate state prediction data of the nuclear reactor at the same sampling moment.

[0103] In one embodiment, Figure 5 As shown, S380 includes:

[0104] S382, obtaining the confidence corresponding to the state prediction data of the nuclear reactor at different sampling times;

[0105] S384, acquiring state data of particles in the nuclear reactor at different sampling times according to the state prediction data and the confidence level corresponding to the state prediction data;

[0106] S386, obtaining estimated state measurement data of the nuclear reactor at different sampling times and confidence levels corresponding to the estimated state measurement data according to the state data;

[0107] S388, obtaining actual state measurement data of the nuclear reactor at different sampling times;

[0108] S390, obtaining residual coefficients of the state measurement data of the nuclear reactor at different sampling times according to the actual state measurement data and the estimated state measurement data;

[0109] S392, obtaining a gain coefficient of the state measurement data according to the confidence level corresponding to the estimated state measurement data;

[0110] S394, according to the gain coefficient and the residual coefficient, the state prediction data and the state measurement data are merged to obtain the optimal estimation data of the state of the nuclear reactor at different sampling times.

[0111] Among them, the estimated state measurement data is a data of the nuclear reactor state measured by the detector calculated by the relevant formula, which is used to estimate the state measurement data of the detector; the actual state measurement data is the real state measurement data actually measured by the detector; the residual coefficient is a coefficient used to characterize the error between the estimated state measurement data and the actual state measurement data. Specifically, according to the state prediction data obtained in the previous embodiment The confidence level corresponding to the state prediction data can be calculated by the following formula:

[0112]

[0113] After obtaining the state prediction data of the nuclear reactor at the sampling time and the corresponding confidence of the state prediction data, N particles are extracted from the nuclear reactor through Gaussian sampling to obtain the state data of these N particles at the sampling time. Then the obtained By mapping the state space to the measurement space, we get Where h is the observation function of the detector. The state data of the nuclear reactor at sampling time t+1 measured by the detector is estimated by the following formula, that is, the estimated state measurement data And the confidence S t+1 :

[0114]

[0115]

[0116] in, R mes It is the measurement uncertainty of the detector. Different types of detectors or the same type of detectors deployed in different locations may lead to changes in the measurement accuracy of different detectors, so there is a certain measurement uncertainty. Obtain the actual state measurement data of the nuclear reactor at the sampling time actually measured by the detector By obtaining the actual state measurement data The estimated state measurement data calculated Compare and get the residual of state measurement data The confidence level corresponding to the estimated state measurement data Calculate the gain factor of the state measurement data: According to the gain coefficient and residual coefficient of the state measurement data, the fusion of the state prediction data and the state measurement data can be realized to obtain the optimal state estimation data s of the nuclear reactor at the sampling time. t+1 And the confidence level P corresponding to the best estimated state data t+1 , and s t+1 With P t+1 The correction can be made based on the residual coefficient and the gain coefficient, namely:

[0117]

[0118]

[0119] In this embodiment, the gain coefficient of the state measurement data is obtained according to the residual coefficient of the state measurement data obtained by comparing the estimated state measurement data of the nuclear reactor with the actual state measurement data and the confidence level corresponding to the estimated state measurement data, and the state best estimation data s t+1 And the confidence level P corresponding to the best estimated state data t+1 Correction can be performed based on the residual coefficient and the gain coefficient, and the accuracy of the best state estimation data and its confidence can be guaranteed when the measured state data changes irregularly, thereby ensuring that an accurate fault classification result can be obtained in the end.

[0120] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0121] Based on the same inventive concept, the embodiment of the present application also provides a nuclear reactor fault classification device for implementing the nuclear reactor fault classification method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more embodiments of the nuclear reactor fault classification device provided below can refer to the limitations of the nuclear reactor fault classification method above, and will not be repeated here.

[0122] In one embodiment, Figure 6As shown, a nuclear reactor fault classification device is provided, comprising: a data acquisition module 100, a data processing module 200 and a fault classification module 300, wherein:

[0123] A data acquisition module 100 is used to acquire the best estimated state data of the nuclear reactor at different sampling times, where the best estimated state data includes data of multiple dimensions;

[0124] The data processing module 200 generates a state best estimation sequence of multiple dimensions based on a preset period according to the state best estimation data; and forms a state estimation matrix according to the state best estimation sequence of multiple dimensions;

[0125] The fault classification module 300 is used to input the state estimation matrix into a preset fault classification model to obtain the state fault classification result and the confidence level corresponding to the state fault classification result; input the state fault classification result and the confidence level corresponding to the state fault classification result into a preset self-learning model to obtain the fault classification result of the nuclear reactor and the confidence level corresponding to the fault classification result.

[0126] The above-mentioned nuclear reactor fault classification device, on the one hand, can obtain accurate fault classification results by obtaining the best estimation data of the state of the nuclear reactor at different sampling times and integrating the data at different sampling times, thereby greatly improving the accuracy of fault classification diagnosis; on the second hand, the state estimation matrix is ​​first input into a preset fault classification model to obtain the state fault classification result and the confidence corresponding to the fault classification result, and then the obtained state fault classification result and the confidence corresponding to the state fault classification result are input into a preset self-learning model, which aggregates different fault classification results and can ensure the accuracy of the final output nuclear reactor fault classification result and the confidence corresponding to the fault classification result, thereby obtaining a high-precision nuclear reactor fault classification result.

[0127] In one embodiment, the data acquisition module 100 is also used to obtain the state data of the nuclear reactor at different sampling times; the state data at different sampling times are fused through the state optimal estimation fusion algorithm to obtain the state optimal estimation data of the nuclear reactor at different sampling times.

[0128] In one embodiment, the data acquisition module 100 is also used to obtain the best estimation data of the state of the nuclear reactor at the previous moment at different sampling moments and the confidence level corresponding to the best estimation data of the state at the previous moment; obtain the state prediction data of the nuclear reactor at different sampling moments according to the best estimation data of the state at the previous moment and the confidence level corresponding to the best estimation data of the state at the previous moment; obtain the state measurement data of the nuclear reactor at different sampling moments; and fuse the state prediction data with the state measurement data through a state optimal estimation fusion algorithm to obtain the best estimation data of the state of the nuclear reactor at different sampling moments.

[0129] In one embodiment, the data acquisition module 100 is also used to obtain the state data of particles in the nuclear reactor at different sampling times at the previous moment based on the best state estimation data at the previous moment and the confidence level of the best state estimation data; input the state data of particles at different sampling times at the previous moment into a preset simulation prediction model, output the state prediction data of particles at different sampling times; and obtain the state prediction data of the nuclear reactor at different sampling times based on the state prediction data of particles at different sampling times.

[0130] In one embodiment, the data acquisition module 100 is also used to obtain the confidence corresponding to the state prediction data of the nuclear reactor at different sampling times; obtain the state data of particles in the nuclear reactor at different sampling times based on the state prediction data and the confidence corresponding to the state prediction data; obtain the estimated state measurement data of the nuclear reactor at different sampling times and the confidence corresponding to the estimated state measurement data based on the state data; obtain the actual state measurement data of the nuclear reactor at different sampling times; obtain the residual coefficient of the state measurement data of the nuclear reactor at different sampling times based on the actual state measurement data and the estimated state measurement data; obtain the gain coefficient of the state measurement data based on the confidence corresponding to the estimated state measurement data. According to the gain coefficient and the residual coefficient, the state prediction data and the state measurement data are fused to obtain the optimal estimated data of the state of the nuclear reactor at different sampling times.

[0131] Each module in the above-mentioned nuclear reactor fault classification device can be implemented in whole or in part by software, hardware and a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each of the above modules.

[0132] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 7 As shown. The computer device includes a processor, a memory and a network interface connected through a system bus. Among them, the processor of the computer device is used 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 the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store nuclear reactor fault classification data. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a nuclear reactor fault classification method is implemented.

[0133] Those skilled in the art will understand that Figure 7 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0134] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0135] The optimal state estimation data of the nuclear reactor at different sampling times are obtained, and the optimal state estimation data include data of multiple dimensions; based on the optimal state estimation data, a multi-dimensional optimal state estimation sequence is generated based on a preset period; based on the multi-dimensional optimal state estimation sequence, a state estimation matrix is ​​constructed; the state estimation matrix is ​​input into a preset fault classification model to obtain a state fault classification result and a confidence level corresponding to the state fault classification result; the state fault classification result and the confidence level corresponding to the state fault classification result are input into a preset self-learning model to obtain a fault classification result of the nuclear reactor and a confidence level corresponding to the fault classification result.

[0136] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0137] The state data of the nuclear reactor at different sampling times are obtained; the state data at different sampling times are fused through the state optimal estimation fusion algorithm to obtain the state optimal estimation data of the nuclear reactor at different sampling times.

[0138] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0139] The best estimated state data of the nuclear reactor at different sampling moments before and the confidence level corresponding to the best estimated state data at the previous moment are obtained; the state prediction data of the nuclear reactor at different sampling moments are obtained according to the best estimated state data at the previous moment and the confidence level corresponding to the best estimated state data at the previous moment; the state measurement data of the nuclear reactor at different sampling moments are obtained; the state prediction data is fused with the state measurement data through the state optimal estimation fusion algorithm to obtain the best estimated state data of the nuclear reactor at different sampling moments.

[0140] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0141] According to the state best estimation data at the previous moment and the confidence of the state best estimation data, the state data of particles in the nuclear reactor at different sampling moments at the previous moment are obtained; the state data of particles at different sampling moments at the previous moment are input into a preset simulation prediction model, and the state prediction data of particles at different sampling moments are output; according to the state prediction data of particles at different sampling moments, the state prediction data of the nuclear reactor at different sampling moments are obtained.

[0142] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0143] Obtain confidences corresponding to state prediction data of a nuclear reactor at different sampling moments; obtain state data of particles in the nuclear reactor at different sampling moments based on the state prediction data and the confidences corresponding to the state prediction data; obtain estimated state measurement data of the nuclear reactor at different sampling moments and the confidences corresponding to the estimated state measurement data based on the state data; obtain actual state measurement data of the nuclear reactor at different sampling moments; obtain residual coefficients of the state measurement data of the nuclear reactor at different sampling moments based on the actual state measurement data and the estimated state measurement data; obtain gain coefficients of the state measurement data based on the confidences corresponding to the estimated state measurement data; fuse the state prediction data with the state measurement data based on the gain coefficient and the residual coefficient to obtain optimal state estimation data of the nuclear reactor at different sampling moments.

[0144] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0145] The optimal state estimation data of the nuclear reactor at different sampling times are obtained, and the optimal state estimation data include data of multiple dimensions; based on the optimal state estimation data, a multi-dimensional optimal state estimation sequence is generated based on a preset period; based on the multi-dimensional optimal state estimation sequence, a state estimation matrix is ​​constructed; the state estimation matrix is ​​input into a preset fault classification model to obtain a state fault classification result and a confidence level corresponding to the state fault classification result; the state fault classification result and the confidence level corresponding to the state fault classification result are input into a preset self-learning model to obtain a fault classification result of the nuclear reactor and a confidence level corresponding to the fault classification result.

[0146] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0147] The state data of the nuclear reactor at different sampling times are obtained; the state data at different sampling times are fused through the state optimal estimation fusion algorithm to obtain the state optimal estimation data of the nuclear reactor at different sampling times.

[0148] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0149] The best estimation data of the state of the nuclear reactor at different sampling moments before and the confidence level corresponding to the best estimation data of the state at the previous moment are obtained; the state prediction data of the nuclear reactor at different sampling moments are obtained according to the best estimation data of the state at the previous moment and the confidence level corresponding to the best estimation data of the state at the previous moment; the state measurement data of the nuclear reactor at different sampling moments are obtained; and the state prediction data and the state measurement data are fused by the state optimal estimation fusion algorithm to obtain the best estimation data of the state of the nuclear reactor at different sampling moments.

[0150] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0151] According to the state best estimation data at the previous moment and the confidence of the state best estimation data, the state data of particles in the nuclear reactor at different sampling moments at the previous moment are obtained; the state data of particles at different sampling moments at the previous moment are input into a preset simulation prediction model, and the state prediction data of particles at different sampling moments are output; according to the state prediction data of particles at different sampling moments, the state prediction data of the nuclear reactor at different sampling moments are obtained.

[0152] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0153] Obtain confidences corresponding to state prediction data of a nuclear reactor at different sampling moments; obtain state data of particles in the nuclear reactor at different sampling moments based on the state prediction data and the confidences corresponding to the state prediction data; obtain estimated state measurement data of the nuclear reactor at different sampling moments and the confidences corresponding to the estimated state measurement data based on the state data; obtain actual state measurement data of the nuclear reactor at different sampling moments; obtain residual coefficients of the state measurement data of the nuclear reactor at different sampling moments based on the actual state measurement data and the estimated state measurement data; obtain gain coefficients of the state measurement data based on the confidences corresponding to the estimated state measurement data; fuse the state prediction data with the state measurement data based on the gain coefficients and the residual coefficients to obtain optimal state estimation data of the nuclear reactor at different sampling moments.

[0154] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0155] The optimal state estimation data of the nuclear reactor at different sampling times are obtained, and the optimal state estimation data include data of multiple dimensions; based on the optimal state estimation data, a multi-dimensional optimal state estimation sequence is generated based on a preset period; based on the multi-dimensional optimal state estimation sequence, a state estimation matrix is ​​constructed; the state estimation matrix is ​​input into a preset fault classification model to obtain a state fault classification result and a confidence level corresponding to the state fault classification result; the state fault classification result and the confidence level corresponding to the state fault classification result are input into a preset self-learning model to obtain a fault classification result of the nuclear reactor and a confidence level corresponding to the fault classification result.

[0156] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0157] The state data of the nuclear reactor at different sampling times are obtained; the state data at different sampling times are fused through the state optimal estimation fusion algorithm to obtain the state optimal estimation data of the nuclear reactor at different sampling times.

[0158] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0159] The best estimation data of the state of the nuclear reactor at different sampling moments before and the confidence level corresponding to the best estimation data of the state at the previous moment are obtained; the state prediction data of the nuclear reactor at different sampling moments are obtained according to the best estimation data of the state at the previous moment and the confidence level corresponding to the best estimation data of the state at the previous moment; the state measurement data of the nuclear reactor at different sampling moments are obtained; and the state prediction data and the state measurement data are fused by the state optimal estimation fusion algorithm to obtain the best estimation data of the state of the nuclear reactor at different sampling moments.

[0160] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0161] According to the state best estimation data at the previous moment and the confidence of the state best estimation data, the state data of particles in the nuclear reactor at different sampling moments at the previous moment are obtained; the state data of particles at different sampling moments at the previous moment are input into a preset simulation prediction model, and the state prediction data of particles at different sampling moments are output; according to the state prediction data of particles at different sampling moments, the state prediction data of the nuclear reactor at different sampling moments are obtained.

[0162] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0163] Obtain confidences corresponding to state prediction data of a nuclear reactor at different sampling moments; obtain state data of particles in the nuclear reactor at different sampling moments based on the state prediction data and the confidences corresponding to the state prediction data; obtain estimated state measurement data of the nuclear reactor at different sampling moments and the confidences corresponding to the estimated state measurement data based on the state data; obtain actual state measurement data of the nuclear reactor at different sampling moments; obtain residual coefficients of the state measurement data of the nuclear reactor at different sampling moments based on the actual state measurement data and the estimated state measurement data; obtain gain coefficients of the state measurement data based on the confidences corresponding to the estimated state measurement data; fuse the state prediction data with the state measurement data based on the gain coefficients and the residual coefficients to obtain optimal state estimation data of the nuclear reactor at different sampling moments.

[0164] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0165] 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, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the 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 (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0166] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0167] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A nuclear reactor fault classification method, It is characterized in that The method comprises: Acquiring state data of the nuclear reactor at different sampling times; wherein the state data includes state measurement data; Obtaining the best estimated data of the state of the nuclear reactor at a previous moment at different sampling moments and the confidence level corresponding to the best estimated data of the state at a previous moment; Obtaining the state prediction data of the nuclear reactor at different sampling moments according to the best estimated state data at the previous moment and the confidence level corresponding to the best estimated state data at the previous moment; The state prediction data and the state measurement data are fused by a state optimal estimation fusion algorithm to obtain the state optimal estimation data of the nuclear reactor at different sampling times, wherein the state optimal estimation data includes data of multiple dimensions; Generate a state best estimation sequence of multiple dimensions based on a preset period according to the state best estimation data; Constructing a state estimation matrix according to the state best estimation sequences of the multiple dimensions; Inputting the state estimation matrix into a preset fault classification model to obtain a state fault classification result and a confidence level corresponding to the state fault classification result; The state fault classification result and the confidence level corresponding to the state fault classification result are input into a preset self-learning model to obtain the fault classification result of the nuclear reactor and the confidence level corresponding to the fault classification result.

2. The method according to claim 1, It is characterized in that The step of obtaining the state prediction data of the nuclear reactor at different sampling times according to the state best estimation data at the previous moment and the confidence level corresponding to the state best estimation data at the previous moment comprises: Acquiring the state data of particles in the nuclear reactor at different sampling moments at the previous moment according to the state best estimation data at the previous moment and the confidence level of the state best estimation data; Inputting the state data of the particle at the previous moment of different sampling moments into a preset simulation prediction model, and outputting the state prediction data of the particle at different sampling moments; The state prediction data of the nuclear reactor at different sampling times are obtained according to the state prediction data of the particles at different sampling times.

3. The method according to claim 2, It is characterized in that The obtaining, according to the state prediction data of the particles at different sampling times, the state prediction data of the nuclear reactor at different sampling times comprises: The following formula is used to determine the state prediction data of the nuclear reactor at different sampling times: Wherein, t represents the moment before the sampling moment; N represents the number of particles in the nuclear reactor; i represents the i-th particle; represents the state prediction data of particle i at sampling time t+1; Represents the state prediction data of the nuclear reactor at sampling time t+1.

4. The method according to claim 1, It is characterized in that The state prediction data and the state measurement data are fused by the state optimal estimation fusion algorithm to obtain the state optimal estimation data of the nuclear reactor at different sampling times, including: Obtaining the confidence corresponding to the state prediction data of the nuclear reactor at different sampling times; Acquire the state data of particles in the nuclear reactor at different sampling times according to the state prediction data and the confidence level corresponding to the state prediction data; Obtaining estimated state measurement data of the nuclear reactor at different sampling times and confidence levels corresponding to the estimated state measurement data according to the state data; Acquire actual state measurement data of the nuclear reactor at different sampling times; Obtaining residual coefficients of the state measurement data of the nuclear reactor at different sampling times according to the actual state measurement data and the estimated state measurement data; Obtaining a gain coefficient of the state measurement data according to the confidence level corresponding to the estimated state measurement data; The state prediction data and the state measurement data are fused according to the gain coefficient and the residual coefficient to obtain the optimal estimation data of the state of the nuclear reactor at the different sampling moments.

5. The method according to claim 4, It is characterized in that The obtaining of the confidence level corresponding to the state prediction data of the nuclear reactor at different sampling times includes: The confidence level corresponding to the state prediction data of the nuclear reactor at different sampling times is determined by the following formula: Wherein, t represents the sampling time; N represents the number of particles in the nuclear reactor; i represents the i-th particle; represents the state prediction data of the nuclear reactor at sampling time t+1; Represents the state data of particle i at sampling time t+1; Represents the confidence level corresponding to the state prediction data of the nuclear reactor at sampling time t+1.

6. A nuclear reactor fault classification device, It is characterized in that The device comprises: A data acquisition module is used to acquire the best estimated state data of the nuclear reactor at different sampling times, and the best estimated state data includes data of multiple dimensions; A data processing module generates a state best estimation sequence of multiple dimensions based on a preset period according to the state best estimation data; and forms a state estimation matrix according to the state best estimation sequence of multiple dimensions; A fault classification module is used to input the state estimation matrix into a preset fault classification model to obtain a state fault classification result and a confidence level corresponding to the state fault classification result; input the state fault classification result and the confidence level corresponding to the state fault classification result into a preset self-learning model to obtain a fault classification result of the nuclear reactor and a confidence level corresponding to the fault classification result; Among them, the data acquisition module is also used to obtain the state measurement data of the nuclear reactor at different sampling times; obtain the best estimation data of the state of the nuclear reactor at the previous moment before the different sampling times and the confidence corresponding to the best estimation data of the state at the previous moment; obtain the state prediction data of the nuclear reactor at different sampling times according to the best estimation data of the state at the previous moment and the confidence corresponding to the best estimation data of the state at the previous moment; obtain the state measurement data of the nuclear reactor at different sampling times; and fuse the state prediction data with the state measurement data through the state optimal estimation fusion algorithm to obtain the best estimation data of the state of the nuclear reactor at different sampling times.

Images