Method and system for determining a nuclear power plant core shroud life prediction model
Patent Information
- Application Number
- CN202610786875.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-06-03
- Publication Date
- 2026-09-01
AI Technical Summary
[0004]本公开要解决的技术问题是为了克服现有技术中核电厂卡件失效数据缺失，无法准确选择合适的寿命预测模型对卡件剩余寿命进行准确预测的缺陷，提供一种核电厂卡件寿命预测模型的确定方法以及系统
[0041] Based on common knowledge in the field, the above-mentioned preferred conditions can be combined arbitrarily to obtain various preferred embodiments of this disclosure.
Smart Images
Figure CN122674497A_ABST
Abstract
Description
Technical Field
[0001] This disclosure relates to the field of instrumentation and control equipment technology in nuclear power plants, and in particular to a method and system for determining a life prediction model for nuclear power plant components. Background Technology
[0002] With the widespread application of digital instrumentation and control systems in nuclear power plants, the problems of instrumentation and control equipment failure and component aging have become increasingly prominent. Frequent electronic card failures have triggered nuclear power plant trips or shutdowns, seriously affecting the safety of nuclear power plants. Statistics show that approximately 50% of nuclear power plant trips or shutdowns are caused by instrumentation and control equipment failures. In the five years following 2006, there were 112 shutdowns and load shedding incidents caused by instrumentation and control circuit board cards, resulting in significant power generation losses.
[0003] Currently, card life analysis faces many challenges. On the one hand, card failure data is scarce, with mainly non-failure data, making it difficult to directly apply machine learning methods that rely on large amounts of failure data. On the other hand, existing life analysis methods are insufficient in accuracy and reliability, unable to accurately predict card life, and thus failing to meet the needs of preventive maintenance and safety management in nuclear power plants. Summary of the Invention
[0004] The technical problem to be solved by this disclosure is to overcome the shortcomings of the prior art in which there is a lack of failure data for nuclear power plant card components, making it impossible to accurately select a suitable life prediction model to accurately predict the remaining life of the card components. This disclosure provides a method and system for determining the life prediction model for nuclear power plant card components.
[0005] This disclosure solves the above-mentioned technical problems through the following technical solution:
[0006] Firstly, a method for determining a nuclear power plant component life prediction model is provided, the method comprising the following steps:
[0007] Obtain the running dataset of the target card, wherein the running dataset includes the output signal values of the target card at multiple consecutive time points;
[0008] The running dataset is input into multiple different candidate prediction models, and the prediction result set corresponding to each candidate prediction model is output; wherein, the prediction result set includes the prediction output signal value of the next time point corresponding to each time point in the running dataset;
[0009] The prediction error of each candidate prediction model is calculated based on the prediction result set;
[0010] The candidate prediction model with the smallest prediction error is determined as the target prediction model.
[0011] Optionally, the multiple different candidate prediction models include at least two of the following:
[0012] Grey prediction model, Kalman filter model, autoregressive model, time series model, linear performance degradation model, support vector regression model, higher-order regression model, and stochastic process prediction model.
[0013] Optionally, the prediction error is the average prediction error, and the step of calculating the prediction error of each candidate prediction model based on the prediction result set specifically includes:
[0014] A set of output signal values is randomly selected from the running dataset as a validation set;
[0015] Calculate the average prediction error of the predicted output signal values at the same time point between the validation set and the prediction result set.
[0016] Optionally, the prediction error is an information criterion value, and the step of calculating the prediction error of each candidate prediction model based on the prediction result set specifically includes:
[0017] Calculate the log-likelihood value for each candidate prediction model based on the prediction result set;
[0018] The information criterion value of the candidate prediction model is calculated based on the log-likelihood value and the corresponding number of parameters of the candidate prediction model. The information criterion value includes the AIC value and / or the BIC value.
[0019] Optionally, the step of obtaining the runtime dataset of the target card may include, prior to:
[0020] Acquire multiple actual output signal values of the target card within a preset operating time;
[0021] In response to the detection of an outlier in the actual output signal value, the outlier is replaced with the average of the actual output signal values at the time points before and after the outlier.
[0022] Optionally, the determination method further includes: determining the duration for which the output signal value decreases to a failure threshold as the remaining lifespan of the target card.
[0023] Secondly, a system for determining a nuclear power plant component life prediction model is provided, the system comprising:
[0024] The acquisition module is used to acquire the running dataset of the target card, which includes the output signal values of the target card at multiple consecutive time points;
[0025] The multi-model prediction module is used to input the running dataset into multiple different candidate prediction models and output a prediction result set corresponding to each candidate prediction model; wherein, the prediction result set includes the prediction output signal value of the next time point corresponding to each time point in the running dataset;
[0026] The calculation module is used to calculate the prediction error of each candidate prediction model based on the prediction result set;
[0027] The determination module is used to determine the candidate prediction model with the smallest prediction error as the target prediction model.
[0028] Optionally, the prediction error is the average prediction error, and the calculation module specifically includes:
[0029] An extraction unit is used to randomly extract a set of output signal values from the running dataset as a verification set;
[0030] The first calculation unit is used to calculate the average prediction error of the predicted output signal values at the same time point between the validation set and the prediction result set.
[0031] Optionally, the prediction error is an information criterion value, and the calculation module specifically includes:
[0032] The second calculation unit is used to calculate the log likelihood value corresponding to each candidate prediction model based on the prediction result set.
[0033] The third calculation unit is used to calculate the information criterion value of the candidate prediction model based on the log-likelihood value and the corresponding number of parameters of the candidate prediction model. The information criterion value includes the AIC value and / or the BIC value.
[0034] Optionally, the determining system further includes:
[0035] The pre-acquisition module is used to acquire multiple actual output signal values of the target card within a preset operating time.
[0036] An abnormal data processing module is used to replace the abnormal value with the average of the actual output signal values at the time points before and after the abnormal value when an abnormal value is detected in the actual output signal value.
[0037] Optionally, the determining system further includes: a remaining life determination module, used to determine the duration for which the output signal value decreases to the failure threshold as the remaining life of the target card.
[0038] Thirdly, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and used to run on the processor, wherein the processor executes the computer program to implement the method for determining the nuclear power plant card life prediction model as described in the first aspect.
[0039] Fourthly, a computer-readable storage medium is provided, on which a computer program is stored, wherein when the computer program is executed by a processor, it implements the method for determining the nuclear power plant card life prediction model as described in the first aspect.
[0040] Fifthly, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the method for determining the nuclear power plant card life prediction model as described in the first aspect.
[0041] Based on common knowledge in the field, the above-mentioned preferred conditions can be combined arbitrarily to obtain various preferred embodiments of this disclosure.
[0042] The positive and progressive effects of this disclosure are as follows: To address the problem in the prior art of missing failure data for nuclear power plant card components, which makes it impossible to accurately select a suitable life prediction model to accurately predict the remaining life of the card components, this disclosure obtains an operational dataset representing the time series of the target card component, and inputs the operational dataset into multiple different candidate prediction models, outputting a prediction result set corresponding to each candidate prediction model, and obtaining the predicted output signal value of the next time point corresponding to each time point in the operational dataset; thereby calculating the prediction error of each candidate prediction model based on the prediction result set, and determining the candidate prediction model with the smallest prediction error as the target prediction model; applying the evaluated and selected target prediction model to the card component life prediction, and obtaining the remaining life prediction result of the card component, which is beneficial for providing a decision-making basis for nuclear power plants to formulate card component maintenance plans. Attached Figure Description
[0043] Figure 1 A flowchart illustrating a method for determining a nuclear power plant card life prediction model, as provided in Embodiment 1 of this disclosure;
[0044] Figure 2 A detailed flowchart of step S13 provided in Embodiment 1 of this disclosure;
[0045] Figure 3 A detailed flowchart of step S13 provided in Embodiment 1 of this disclosure;
[0046] Figure 4 This is a partial flowchart of a method for determining a nuclear power plant card life prediction model according to Embodiment 1 of this disclosure;
[0047] Figure 5A detailed flowchart of a method for determining a nuclear power plant card life prediction model provided in Embodiment 1 of this disclosure;
[0048] Figure 6 This is a schematic diagram of a system for determining a nuclear power plant card life prediction model, provided in Embodiment 2 of this disclosure.
[0049] Figure 7 This is a schematic diagram of the structure of an electronic device provided in Embodiment 3 of this disclosure. Detailed Implementation
[0050] The present disclosure is further illustrated below by way of embodiments, but the present disclosure is not limited to the scope of the embodiments described herein.
[0051] The prefixes such as "first" and "second" used in this disclosure are merely for distinguishing different descriptive objects and do not limit the position, order, priority, quantity, or content of the described objects. The use of ordinal numbers and other prefixes used to distinguish descriptive objects in this disclosure does not constitute a limitation on the described objects. The description of the described objects is given in the claims or the context of the embodiments, and should not be construed as an unnecessary limitation. Furthermore, in the description of this embodiment, unless otherwise stated, "multiple" means two or more.
[0052] Example 1
[0053] Figure 1 This is a flowchart of a method for determining a nuclear power plant component life prediction model provided in Embodiment 1. The method includes the following steps:
[0054] S11. Obtain the running dataset of the target card, wherein the running dataset includes the output signal values of the target card at multiple consecutive time points.
[0055] In this embodiment, the output signal values obtained by the target card in the high-temperature constant acceleration test can be used as a component of the running dataset. That is, the running dataset can be a time series of output signal values. The output signal values can be current or voltage, etc.
[0056] S12. Input the running dataset into multiple different candidate prediction models respectively, and output the prediction result set corresponding to each candidate prediction model; wherein, the prediction result set includes the prediction output signal value of the next time point corresponding to each time point in the running dataset.
[0057] In this embodiment, multiple different statistical models are selected as candidate prediction models. Each statistical model is based on different principles and can capture data features from different perspectives. Therefore, using multiple statistical models to analyze and predict the lifespan of target components in nuclear power plants can improve the accuracy and reliability of the prediction.
[0058] S13. Calculate the prediction error of each candidate prediction model based on the prediction result set.
[0059] In this embodiment, the prediction error of each candidate prediction model is calculated, and a model suitable for predicting the lifespan of the target card can be selected, providing a scientific basis for preventive maintenance and safety management of nuclear power plants.
[0060] S14. The candidate prediction model with the smallest prediction error is determined as the target prediction model.
[0061] In this embodiment, the candidate prediction model with the smallest prediction error is ultimately determined as the target prediction model, effectively solving the problem of predicting the remaining life of nuclear power plant components, improving the accuracy and reliability of component life prediction, and overcoming the model misfit problem caused by missing data. During the comparison and selection process of different candidate prediction models, the target prediction model that best matches the actual situation is selected. Using this target prediction model in subsequent component remaining life predictions makes the corresponding prediction results closer to the actual remaining life. Therefore, through accurate life prediction, nuclear power plants can optimize component maintenance strategies, schedule maintenance work in advance, reduce the frequency of reactor shutdowns and equipment replacements, reduce power generation losses, improve the overall efficiency and safety of nuclear power plants, promote the development of nuclear power plants towards intelligent operation and maintenance, and enhance the reliability and stability of nuclear power systems.
[0062] In one alternative implementation, the plurality of different candidate prediction models includes at least two of the following:
[0063] Grey prediction model, Kalman filter model, autoregressive model, time series model, linear performance degradation model, support vector regression model, higher-order regression model, and stochastic process prediction model.
[0064] In this implementation, different candidate prediction models are based on different principles and can capture data features from different perspectives. Specifically, the grey prediction model is suitable for a small amount of data and can predict future trends by generating and processing the original data to establish a differential equation model; the Kalman filter model uses the linear system state equation to fuse predicted and observed values to achieve the optimal estimate of the system state; the support vector regression model aims to minimize the difference between predicted and true values while maximizing the margin, thereby improving the model's generalization ability.
[0065] In one specific implementation, the candidate prediction model includes a grey prediction model. Taking the grey prediction model GM(1,1) as an example, first, the output signal values in the running dataset are subjected to a level ratio test to determine whether they satisfy the quasi-exponential law. Then, a new data column is generated through a single accumulation, a data matrix is constructed to calculate parameters, and a first-order differential equation is established and solved to obtain the prediction model. The input data of the grey prediction model is: The output expression of the grey prediction model is: Where a represents the development coefficient and b represents the grayscale effect; Let k = 1, 2, ..., n, representing the k-th output signal value in the unprocessed running dataset, i.e., the k-th data point in the original data sequence. When the next output signal value is predicted... Then, the earliest one will be removed. The sequence becomes At this time, the original It then becomes the first data point in the new sequence.
[0066] In one specific implementation, the candidate prediction model includes a Kalman filter model, and the input data is: ,in Let k = 1, 2, ..., n, representing the k-th output signal value in the unaccumulated running dataset. The state update equation is: in This represents the updated state estimate at time t, i.e., the predicted output signal value at time t. z represents the state estimate at the previous time t-1; t is the actual observed value at time t, i.e., the actual output signal value at time t. Kt is the Kalman gain, which determines how new observation data is incorporated into the state estimation. Its value is between 0 and 1, and this coefficient is used to balance the contributions of the predicted and observed values to the final estimation result. Ht is the observation matrix, used to map the system's state vector to the observation space, representing the transformation relationship from the system state to the observed data. Its dimension and structure depend on the characteristics of the system state and the observed data.
[0067] In one specific implementation, the candidate prediction model includes an autoregressive model, the input of which is... X k Let k = 1, 2, ..., t-1, representing the k-th output signal value in the unaccumulated running dataset. The output expression of the autoregressive model is: Where α1, ..., α n u is a constant coefficient t This represents the random disturbance (noise) term. The autoregressive model calculates the t-th predicted output signal value based on the first t-1 output signal values.
[0068] In one specific implementation, the candidate prediction model includes a time series model, which comprises three parts: an autoregressive model, a differencing process, and a moving average model. The input to the time series model is... , representing t-1 consecutive output signal values, the output expression of the time series model is: Where Y t Let t represent the predicted output signal value, p be the autoregression order, and q be the moving average order. It is a time-stationary series after differencing. c is a constant term, a i These are the parameters of the AR (Autoregressive) model, used to describe the relationship between the current value and the values at p past time points; b j These are parameters of the MA (Moving Average) model, used to describe the relationship between the current value and the error over the past q time points.
[0069] In one specific implementation, the candidate prediction model includes a degradation process model. This degradation process model includes a linear degradation model. For the same device, based on n sets of complete historical data, n sets of initial performance and degradation rates of the system are obtained through linear regression. The variance estimate of the random noise following a normal distribution is calculated using the first-moment estimation method. Then, the first-moment estimation method is used to calculate the n sets of parameters to determine the model parameters. Finally, the parameters are assigned to the model to achieve device performance prediction. For example, for a linear degradation model, the input needs to be a sequence of historical data corresponding to the usage duration of the target card throughout its complete lifecycle. And the data sequence corresponding to the output signal value of the target card that needs to be predicted. The output expression of this linear degradation model is: Where Y t Let t represent the predicted output signal value, β0 represent the initial performance of the system, β1 represent the degradation rate of the system performance, and ε(t) represent the system noise term, with ε(t) following a normal distribution with expectation of 0 and variance of σ². For another linear degradation model, only the data sequence corresponding to the target card's output signal value needs to be input. The t-th output signal value Y can then be obtained from the output expression. t .
[0070] In one specific implementation, the candidate prediction model includes a support vector regression model, which requires inputting the training set, test set, and prediction set into the support vector regression model for training. The training set is... The test set is The prediction set is The initialization parameters are W is the normal vector of the hyperplane, i.e., the weight of the regression function, and b is the bias term. Thus, the optimal parameters of the model are output. and model prediction results .
[0071] In one specific implementation, the candidate prediction model includes a higher-order regression model, the input of which is... The output expression of the higher-order regression model is: , where β i (i=1,2,3) are the parameters of the higher-order regression model, and ε is the error term. Where X... k k=1,2,…,t-1, can represent the working time of the target card; Y t This can be represented as X representing the working time of the target card. t The output signal value.
[0072] In one specific implementation, the candidate prediction model includes a stochastic process model, the input of which is... , usually expressed as Where s(n) is the original signal and v(n) is the noise signal. The output expression of the stochastic process model is: ;s ^ (n) is the estimated original signal, h opt x(n) is the unit impulse response of the optimal Wiener filter, and x(n) is the noisy observation signal. " indicates convolution operation.
[0073] In one optional implementation, the prediction error is the average prediction error, such as... Figure 2 As shown, step S13 specifically includes:
[0074] S131. Randomly select a set of output signal values from the running dataset as a verification set.
[0075] In this embodiment, a set can be randomly selected from the running dataset as a validation set. For example, a running dataset may contain 100 output signal values, and 30 output signal values may be randomly selected as the validation set. The remaining 70 output signal values may be used as the training set to input into the candidate prediction model for prediction. Alternatively, all 100 output signal values may be used as the training set to input into the candidate prediction model for prediction.
[0076] S132. Calculate the average prediction error of the predicted output signal values at the same time point between the validation set and the prediction result set.
[0077] In one specific implementation, cross-validation can be used to validate different candidate prediction models. For example, the given dataset is split into a training set and a validation set, and their average prediction error is calculated. This process is repeated to obtain new training and validation sets. The training set is then input into each candidate prediction model for prediction, resulting in a corresponding prediction result set. The predicted output signal at the corresponding time point in the prediction result set is compared with the output signal value at the same time point in the validation set, and the average prediction error between the two sets of data is calculated. Finally, the candidate prediction model with the smallest average prediction error is selected as the target prediction model.
[0078] In one optional implementation, the prediction error is an information criterion value, such as... Figure 3 As shown, step S13 specifically includes:
[0079] S133. Calculate the log-likelihood value corresponding to each candidate prediction model based on the prediction result set.
[0080] In this embodiment, the formula for calculating the log-likelihood value is as follows:
[0081]
[0082] Wherein, this formula represents the output signal values x1, x2, ..., x in the prediction result set of each candidate prediction model. n Follows a probability distribution f ( x | θ ),in θ This is an unknown parameter of the candidate prediction model, such as mean, variance, etc. L( θ) This indicates that all output signal values are in this parameter. θ The joint probability of occurrence. For ease of calculation, the logarithm is usually taken to convert it into a log-likelihood value.
[0083] S134. Calculate the information criterion value of the candidate prediction model based on the log-likelihood value and the corresponding number of parameters of the candidate prediction model. The information criterion value includes the AIC value and / or the BIC value.
[0084] In this embodiment, the AIC value can be calculated based on the Akaike Information Criterion (AIC) or the Bayesian Information Criterion (BIC). The principles of the AIC and BIC criteria differ: the AIC criterion selects a good candidate prediction model from a prediction perspective, while the BIC criterion selects a candidate prediction model that best fits the existing data from a fitting perspective.
[0085] In one specific implementation, the formula for calculating the AIC value is: AIC = 2k - 2ln(L).
[0086] Here, k represents the number of parameters, and L is the maximum likelihood estimate of the parameters of the corresponding candidate prediction model. A small k means a simple model, while a large L means a precise model. Therefore, the AIC value balances simplicity and precision when evaluating a model. Assuming a choice is made among n candidate prediction models, the AIC values of all n candidate models can be calculated at once, and the candidate prediction model with the smallest AIC value can be selected as the target prediction model. The AIC method aims to find the model that best explains the data while containing the fewest free parameters.
[0087] In one specific implementation, the formula for calculating the BIC value is: BIC = kln(n) - 2ln(L).
[0088] Where k is the number of model parameters, n is the number of samples, and L is the maximum likelihood estimate of the candidate prediction model parameters. A smaller BIC value indicates a better balance between fitting the data and model complexity. The penalty term kln(n) of BIC increases with model complexity, which helps prevent the model from becoming too complex and avoids overfitting.
[0089] In one alternative implementation, such as Figure 4 As shown, the steps preceding step S11 include:
[0090] S01. Obtain multiple actual output signal values of the target card within a preset operating time.
[0091] In this embodiment, correlation analysis can be performed on the acquired multiple actual output signal values to ensure the reliability of the data. For example, the correlation analysis may include detecting whether there are outliers in the actual output signal values, or detecting whether there are missing values in the actual output signal values that cause multiple actual output signal values to be discontinuous.
[0092] S02. In response to detecting an outlier in the actual output signal value, the outlier is replaced with the average of the actual output signal values at the time points before and after the outlier.
[0093] In this embodiment, outliers among the detected actual output signal values are deleted, and the missing points at the deleted outliers are filled with the data mean to obtain the processed original data, which serves as the running dataset and provides a reliable data foundation for subsequent model construction.
[0094] In one optional implementation, the determining method further includes: determining the duration for which the output signal value decreases to a failure threshold as the remaining lifespan of the target card.
[0095] In this embodiment, the research problem is typically identified, and a target card from a nuclear power plant is selected as the analysis object. The data source is clearly defined as the high-temperature constant-acceleration test data of the target card. This allows for the determination of analytical parameters, with the remaining lifetime defined as the time it takes for the target card's output signal value to decrease to the failure threshold. During normal operation, fluctuations in the input signal will be reflected in the output of the card. Changes in output signal values (such as voltage and current) are usually an important indicator of the card's health status. During the normal phase of the target card's lifespan, the output signal value should remain within the designed range. However, when the internal components of the target card gradually age or are affected by the external environment, the output signal value will deviate from the normal value. If a key parameter of the output signal value reaches or exceeds the designed threshold at a certain moment, it indicates that the target card has failed, meaning its service life has ended. Therefore, the remaining lifetime of the target card can be determined based on the time it takes for the output signal value to decrease to the aging threshold, thereby ensuring the safe operation of the nuclear power plant.
[0096] In one specific implementation method Figure 5 This is a flowchart illustrating the method for determining a specific nuclear power plant component life prediction model.
[0097] Step 1: Define the research question: Select a certain nuclear power plant card as the analysis object, clarify that the data source of the operating dataset is the high temperature constant accelerated test data of the card, and define the remaining life as the time it takes for the signal output value of the target card to decrease to the failure threshold.
[0098] Step 2, Data Analysis and Preprocessing: Correlation analysis is performed on the collected data to identify and remove outliers. The missing data points of the removed outliers are filled using the data mean. The processed raw data is used as the running dataset for the candidate prediction model, providing a reliable data foundation for the subsequent construction of the candidate prediction model.
[0099] Step 3: Establish and fit candidate prediction models: Based on data characteristics and research needs, select multiple candidate prediction models from Kalman filter model, support vector regression model, grey prediction model, and autoregressive model. For each candidate prediction model, operate according to its specific data processing method.
[0100] Step 4, Model Evaluation and Comparison: Use the AIC criterion, BIC criterion, or cross-validation method to evaluate and compare the performance of candidate prediction models.
[0101] Step 5, Model Application: The optimal candidate prediction model selected from the evaluation is applied to the card life prediction to obtain the remaining life prediction results of the target card, providing a decision-making basis for nuclear power plants to formulate card maintenance plans.
[0102] Example 2
[0103] Corresponding to Embodiment 1 of the aforementioned method for determining the life prediction model of nuclear power plant components, this disclosure also provides an embodiment of a system for determining the life prediction model of nuclear power plant components.
[0104] Figure 6 This is a schematic diagram of a system for determining a nuclear power plant component life prediction model, as provided in this embodiment. The determining system 20 includes:
[0105] The acquisition module 201 is used to acquire the running dataset of the target card, the running dataset including the output signal values of the target card at multiple consecutive time points;
[0106] The multi-model prediction module 202 is used to input the running dataset into multiple different candidate prediction models respectively, and output the prediction result set corresponding to each candidate prediction model; wherein, the prediction result set includes the prediction output signal value of the next time point corresponding to each time point in the running dataset;
[0107] The calculation module 203 is used to calculate the prediction error of each candidate prediction model based on the prediction result set;
[0108] The determination module 204 is used to determine the candidate prediction model with the smallest prediction error as the target prediction model.
[0109] In this embodiment, the candidate prediction model with the smallest prediction error is ultimately determined as the target prediction model, effectively solving the problem of predicting the remaining life of nuclear power plant components, improving the accuracy and reliability of component life prediction, and overcoming the model misfit problem caused by missing data. During the comparison and selection process of different candidate prediction models, the target prediction model that best matches the actual situation is selected. Using this target prediction model in subsequent component remaining life predictions makes the corresponding prediction results closer to the actual remaining life. Therefore, through accurate life prediction, nuclear power plants can optimize component maintenance strategies, schedule maintenance work in advance, reduce the frequency of reactor shutdowns and equipment replacements, reduce power generation losses, improve the overall efficiency and safety of nuclear power plants, promote the development of nuclear power plants towards intelligent operation and maintenance, and enhance the reliability and stability of nuclear power systems.
[0110] For the system embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs.
[0111] Example 3
[0112] Figure 7 This is a schematic diagram of the structure of an electronic device shown in this embodiment. The electronic device includes a memory, a processor, and a computer program stored in the memory and used to run on the processor. When the processor executes the computer program, it implements the method for determining the nuclear power plant card life prediction model described in Embodiment 1 above. Figure 7 The electronic device 30 shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments disclosed herein.
[0113] like Figure 7 As shown, the electronic device 30 can be manifested as a general-purpose computing device, such as a server device. The components of the electronic device 30 may include, but are not limited to: at least one processor 31, at least one memory 32, and a bus 33 connecting different system components (including memory 32 and processor 31).
[0114] Bus 33 includes a data bus, an address bus, and a control bus.
[0115] The memory 32 may include volatile memory, such as random access memory (RAM) 321 and / or cache memory 322, and may further include read-only memory (ROM) 323.
[0116] The memory 32 may also include a program tool 325 (or utility) having a set (at least one) program module 324, such program module 324 including but not limited to: an operating system, one or more application programs, other program modules and program data, each or some combination of these examples may include an implementation of a network environment.
[0117] The processor 31 executes various functional applications and data processing by running computer programs stored in the memory 32, such as the method for determining the nuclear power plant card life prediction model provided in Embodiment 1 above.
[0118] Electronic device 30 can also communicate with one or more external devices 34 (e.g., keyboard, pointing device, etc.). This communication can be performed via input / output (I / O) interface 35. Furthermore, electronic device 30 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 36. Figure 7 As shown, network adapter 36 communicates with other modules of electronic device 30 via bus 33. It should be understood that, although... Figure 7 Not shown, other hardware and / or software modules may be used in conjunction with electronic device 30, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID (disk array) systems, tape drives, and data backup storage systems.
[0119] It should be noted that although several units / modules or sub-units / modules of the electronic device have been mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, according to embodiments of this disclosure, the features and functions of two or more units / modules described above can be embodied in one unit / module. Conversely, the features and functions of one unit / module described above can be further divided and embodied by multiple units / modules.
[0120] Example 4
[0121] This disclosure also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the method for determining the nuclear power plant card life prediction model provided in Embodiment 1 above.
[0122] The readable storage medium may be more specifically adopted, including but not limited to: portable disk, hard disk, random access memory, read-only memory, erasable programmable read-only memory, optical storage device, magnetic storage device, or any suitable combination thereof.
[0123] Example 5
[0124] This disclosure also provides a computer program product, including a computer program that, when executed by a processor, implements the method for determining the nuclear power plant card life prediction model as described in Embodiment 1 above.
[0125] The program code for executing the computer program product of this disclosure can be written in any combination of one or more programming languages, and the program code can be executed entirely on a user device, partially on a user device, as a stand-alone software package, partially on a user device and partially on a remote device, or entirely on a remote device.
[0126] While specific embodiments of this disclosure have been described above, those skilled in the art should understand that these are merely illustrative examples, and the scope of protection of this disclosure is defined by the appended claims. Those skilled in the art can make various changes or modifications to these embodiments without departing from the principles and essence of this disclosure, but all such changes and modifications fall within the scope of protection of this disclosure.
Claims
1. A method for determining a nuclear power plant component life prediction model, characterized in that, The determination method includes the following steps: Obtain the running dataset of the target card, wherein the running dataset includes the output signal values of the target card at multiple consecutive time points; The running dataset is input into multiple different candidate prediction models, and the prediction result set corresponding to each candidate prediction model is output; wherein, the prediction result set includes the prediction output signal value of the next time point corresponding to each time point in the running dataset; The prediction error of each candidate prediction model is calculated based on the prediction result set; The candidate prediction model with the smallest prediction error is determined as the target prediction model.
2. The determination method as described in claim 1, characterized in that, Multiple different candidate prediction models include at least two of the following: Grey prediction model, Kalman filter model, autoregressive model, time series model, linear performance degradation model, support vector regression model, higher-order regression model, and stochastic process prediction model.
3. The determination method as described in claim 1, characterized in that, The prediction error is the average prediction error. The step of calculating the prediction error of each candidate prediction model based on the prediction result set specifically includes: A set of output signal values is randomly selected from the running dataset as a validation set; Calculate the average prediction error of the predicted output signal values at the same time point between the validation set and the prediction result set.
4. The determination method as described in claim 1, characterized in that, The prediction error is an information criterion value. The steps for calculating the prediction error of each candidate prediction model based on the prediction result set specifically include: Calculate the log-likelihood value for each candidate prediction model based on the prediction result set; The information criterion value of the candidate prediction model is calculated based on the log-likelihood value and the corresponding number of parameters of the candidate prediction model. The information criterion value includes the AIC value and / or the BIC value.
5. The determination method as described in claim 1, characterized in that, The step of obtaining the runtime dataset of the target card includes, prior to: Acquire multiple actual output signal values of the target card within a preset operating time; In response to the detection of an outlier in the actual output signal value, the outlier is replaced with the average of the actual output signal values at the time points before and after the outlier.
6. The determining method according to any one of claims 1-5, characterized in that, The determination method further includes: determining the remaining lifespan of the target card as the duration during which the output signal value decreases to the failure threshold.
7. A system for determining a nuclear power plant component life prediction model, characterized in that, The determining system includes: The acquisition module is used to acquire the running dataset of the target card, which includes the output signal values of the target card at multiple consecutive time points; The multi-model prediction module is used to input the running dataset into multiple different candidate prediction models and output a prediction result set corresponding to each candidate prediction model; wherein, the prediction result set includes the prediction output signal value of the next time point corresponding to each time point in the running dataset; The calculation module is used to calculate the prediction error of each candidate prediction model based on the prediction result set; The determination module is used to determine the candidate prediction model with the smallest prediction error as the target prediction model.
8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and for running on the processor, characterized in that, When the processor executes the computer program, it implements the method for determining the nuclear power plant card life prediction model as described in 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 the processor, it implements the method for determining the nuclear power plant card life prediction model as described in 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 the processor, it implements the method for determining the nuclear power plant card life prediction model as described in any one of claims 1-6.