A method and device for early warning of events based on nuclear power control logic

Abstract

The present application relates to nuclear power plant event early warning technical field, aims at solving how to detect whether the current operation state is normal in the process of unit operation and how to predict the time of abnormal event problems, and provides an event early warning method and device based on nuclear power control logic. The method comprises the following steps: acquiring real-time data, calculating the real-time data, judging whether an abnormal event will occur, when judging that an abnormal event will occur, using a key working condition parameter trend prediction algorithm based on an adaptive linear model to predict the change trend of the real-time data, to predict the time of the abnormal event, outputting and alarming the time of the abnormal event. The present application can monitor, analyze and early warn abnormal events such as shutdown, shutdown and load rejection in the operation process, can detect whether the current operation state is normal, predict the time of the abnormal event, and further provide auxiliary support for ensuring the stable and safe operation of the power plant.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of nuclear power plant event early warning, and in particular to an event early warning method and device based on nuclear power control logic. BACKGROUND

[0002] During normal operation of a nuclear power plant, unexpected protection actions such as load shedding, shutdown, and reactor shutdown may occur due to equipment failure, instrument control system failure, power grid failure, fire, insufficient personnel skills, human error, and the like.

[0003] Currently, abnormal conditions during unit operation are mainly discovered based on alarm information of a DCS system, and only when a threshold is reached will the corresponding alarm be triggered, lacking analysis of trends. In addition, although most typical abnormal events of a unit have different precursors before leading to an accident, DCS alarms cannot completely cover these events, nor can they timely prompt operators according to fluctuations in parameter relevance, so it is difficult to attract attention in the early stage of abnormal events.

[0004] Traditional monitoring methods are based on control charts of single variables, which have been unable to reflect changes in complex industrial processes. Modern process monitoring is based on data-driven methods, and data-driven metrics are directly derived from process data. Although big data of modern industrial systems brings rich information, engineers and operators cannot empirically derive assessments of process operation from this vast amount of data. SUMMARY

[0005] The main purpose of the present application is to provide an event early warning method and device based on nuclear power control logic, solving the problem of how to detect whether the current operating state is normal during unit operation.

[0006] Another purpose of the present application is to provide an event early warning method and device based on nuclear power control logic, solving the problem of how to predict the time of occurrence of the abnormal event.

[0007] To achieve the above purpose, the present application provides the following technical solutions:

[0008] In a first aspect, the present application provides an event early warning method based on nuclear power control logic, comprising:

[0009] S100, acquiring real-time data;

[0010] S200, calculating the real-time data to determine whether an abnormal event will occur;

[0011] S300, when it is determined that an abnormal event will occur, using a key working condition parameter trend prediction algorithm based on an adaptive linear model to predict the change trend of the real-time data to predict the time of occurrence of the abnormal event;

[0012] S400, output the time of the abnormal event and alarm.

[0013] In some embodiments, the abnormal event in the unit operation is monitored by the abnormal monitoring algorithm based on the multi-modal DPCA, and the abnormal event includes at least one of the shutdown event, the shutdown event and the load rejection event.

[0014] In some embodiments, S200 includes:

[0015] S210, grouping the real-time data to obtain a plurality of groups of data;

[0016] S220, calculating the covariance matrix of the plurality of groups of data by using the abnormal monitoring algorithm based on the multi-modal DPCA;

[0017] S230, calculating the Hotelling statistics and the residual statistics of each group of data based on the covariance matrix by using the abnormal monitoring algorithm based on the multi-modal DPCA;

[0018] S240, judging whether an abnormal event will occur based on the Hotelling statistics and the residual statistics of the plurality of groups of data, and the preset Hotelling statistics threshold and the residual statistics threshold corresponding to the plurality of groups of data.

[0019] In some embodiments, S240 includes:

[0020] S241, obtaining a historical sample data set, and dividing the historical sample data set into a plurality of initial data sets;

[0021] S242, calculating the covariance matrix of the plurality of initial data sets;

[0022] S243, calculating the eigenvalues and eigenvectors of the plurality of covariance matrices;

[0023] S244, sorting the plurality of eigenvalues, screening the plurality of eigenvalues by using the cumulative variance contribution rate, merging the plurality of initial data sets, and generating a plurality of intermediate data sets;

[0024] S245, modeling the plurality of intermediate data sets, obtaining a plurality of PCA models, and calculating the residual between the plurality of PCA models;

[0025] S246, merging the plurality of intermediate data sets based on the preset threshold and the residual between the plurality of PCA models, generating a new plurality of data sets, and

[0026] S247, repeating S242 to S246 until no new plurality of data sets is generated, and executing S248;

[0027] S248, the Hotelling statistic threshold value and the residual statistic threshold value of each group of PCA models are calculated by using the PCA method.

[0028] In some embodiments, the establishment process of the fault detection model comprises:

[0029] (1) the initial data set is represented as and the PCA model is constructed by the following iterative process;

[0030] (2) for each group of the number of retained principal components is determined by cross-validation, and the test data set is part of

[0031] (3) similarity comparison is performed between two PCA models, and the residual error is used as the similarity measure:

[0032]

[0033] (4) if the residual error is less than the threshold value f, the corresponding data groups are merged, and the newly generated data groups are represented as go to step (2), otherwise stop;

[0034] (5) determine the detection threshold value: the threshold value is the maximum residual error of the corresponding modal training set itself.

[0035] It should be noted that steps (1)-(5) described above describe how to model the fault detection model, and this part of the model establishment is offline. The actual monitoring process is to input the online data into the model in time sequence, and the monitoring itself is online. The following is a description of the online fault monitoring method.

[0036] Based on the above multi-modal PCA model, the abnormal detection method based on multi-modal PCA is as follows:

[0037] (1) when receiving online data, calculate the residual error SPE of each group of modalities by formula (1);

[0038] (2) detection logic: if at least one residual error is less than the threshold value, it is a normal sample, otherwise it is a fault sample.

[0039] In some embodiments, S300 comprises:

[0040] S310, when it is judged that an abnormal event will occur, the empirical mode decomposition method is used to denoise and trend extract the real-time data to obtain the preprocessed data;

[0041] ​​S320, input the preprocessed data into the pre-constructed linear model to obtain an estimated value of the target variable;

[0042] S330, based on the alarm threshold of the target variable and the estimated value of the target variable, calculate the time of occurrence of the abnormal event.

[0043] In some embodiments, the key working condition parameter trend prediction algorithm based on an adaptive exponential model comprises:

[0044] (1) collect data z k at time t k with a window length of s; ks , perform empirical mode decomposition on it to extract an IMF component containing the overall trend;

[0045] (2) assign initial values to the parameters of the linear model;

[0046] (3) at time k, based on the data k from time t i to t i , train a linear model f(t s , φ, θ);

[0047] (4) estimate the predicted value of k * to k through the linear model f(t * , φ, θ) as follows:

[0048]

[0049] (5) wait for the next time data to come, and jump to step (1);

[0050] At time k, solve the equation to obtain the optimal solution (φ * , θ k )k, and obtain the estimated value of the target variable at time k:

[0051] z * = φ * + θ k t * ;

[0052] Given the alarm threshold ξ of the target variable, it satisfies:

[0053] ξ = φ * + θ k (t k + T k );

[0054] The remaining shutdown time is obtained:

[0055]

[0056] In the formula, T k represents the time of abnormal event occurrence, ξ represents the alarm threshold of the target variable, and φ * , θ * represents the optimal solution of the random variable at the k moment, that is, the estimated value of the target variable at the k moment.

[0057] In a second aspect, the present application provides an event early warning device based on nuclear power control logic, comprising:

[0058] A data acquisition module is configured to acquire real-time data.

[0059] An abnormality judgment module is configured to calculate the real-time data by using an abnormality monitoring algorithm based on multi-modal DPCA, and determine whether an abnormal event will occur.

[0060] An abnormality prediction module is configured to predict the change trend of the real-time data by using a key working condition parameter trend prediction algorithm based on an adaptive linear model when it is determined that an abnormal event will occur, and predict the time of the abnormal event occurrence.

[0061] An output module is configured to output the time of the abnormal event occurrence and alarm.

[0062] In a third aspect, the present application provides a computer device comprising a memory and a processor, wherein the memory stores computer readable instructions, and the processor implements the steps of the event early warning method based on nuclear power control logic when executing the computer readable instructions.

[0063] In a fourth aspect, the present application provides a computer readable storage medium, wherein the computer readable storage medium stores computer readable instructions, and the computer readable instructions implement the steps of the event early warning method based on nuclear power control logic when executed.

[0064] Compared with the prior art, the event early warning method and device based on nuclear power control logic have the following beneficial effects:

[0065] The present application can monitor, analyze and early warn abnormal events such as shutdown, shutdown and load shedding during operation, can detect whether the current operation state is normal, can predict the time of the abnormal event occurrence, and can further provide auxiliary support for ensuring smooth and safe operation of the power plant. BRIEF DESCRIPTION OF DRAWINGS

[0066] In order to more clearly illustrate the technical solutions of the present application, the following will briefly introduce the drawings needed to be used in the technical description.

[0067] Figure 1 The flowchart of the event early warning method based on nuclear power control logic provided by the present application;

[0068] Figure 2 This is a schematic diagram of the initial grouping provided by the present invention;

[0069] Figure 3 DPCA fault detection flow chart provided by the present invention;

[0070] Figure 4 The overall roadmap for trend prediction of key operating parameters provided by the present invention;

[0071] Figure 5 The EMD flow chart provided by the present invention;

[0072] Figure 6 This is a schematic diagram of the event warning device based on nuclear power control logic provided by the present invention. DETAILED DESCRIPTION

[0073] The following is further detailed description through specific implementation methods.

[0074] like Figure 1 As shown, the present invention provides an event warning method based on nuclear power control logic. Developed for major operational events in nuclear power plants, it can identify abnormal changes in equipment parameters related to various scenarios such as shutdowns, outages, and load shedding, and accurately predict possible events and their expected occurrence times. The warning method specifically includes the following steps:

[0075] S100, obtaining real-time data;

[0076] S200, using an abnormality monitoring algorithm based on multimodal DPCA to calculate real-time data to determine whether an abnormal event will occur, wherein the abnormal event is at least one of a reactor shutdown event, a shutdown event, and a load rejection event;

[0077] S300, when it is determined that an abnormal event will occur, using a key operating condition parameter trend prediction algorithm based on an adaptive linear model to predict the change trend of real-time data to predict the time when the abnormal event will occur;

[0078] S400: Output the time when the abnormal event occurs and generate an alarm.

[0079] By detecting anomalies in event-related parameters and dividing the input parameter range according to different events, the method uses data processing, feature extraction, multivariate statistical analysis, dynamic principal component analysis and other algorithms to conduct research on anomaly detection of shutdown events and four scenarios. Then, trend prediction is introduced to determine how long it will take for an abnormal event to occur. It uses regression modeling based on historical power plant operating data and the correlation between parameters to predict the operating trends of relevant parameters of specific events in the future. Finally, the information is merged and output as early warning information, thereby providing auxiliary support for ensuring the smooth and safe operation of the power plant.

[0080] In one embodiment, in step S100, the real-time data is unit real-time operation data or platform twin system data.

[0081] In one embodiment, in step S200, since the online monitoring data deviates from the behavior pattern under normal working conditions when abnormal working conditions caused by equipment failure or failure of the nuclear power plant occur, the present application judges whether an abnormal event will occur by monitoring the real-time data. Since the nuclear reactor has multiple working modes under normal working conditions, the present application uses an abnormal monitoring algorithm based on multi-modal DPCA to monitor the abnormality.

[0082] In one embodiment, in step S200, the abnormality detection is performed by establishing a model offline and then performing online detection by using the model. Specifically, in the offline modeling stage, a monitoring model of the normal working condition is established by using the historical data under the normal working condition, in the online monitoring stage, the residual signal corresponding to the online data is generated by using the monitoring model to indicate the degree of deviation of the online data from the normal working condition, thereby determining the abnormal event alarm.

[0083] In one embodiment, step S200 specifically comprises:

[0084] S210, grouping the real-time data to obtain a plurality of groups of data;

[0085] S220, calculating the covariance matrix of the plurality of groups of data by using the abnormal monitoring algorithm based on multi-modal DPCA;

[0086] S230, calculating the Hotelling statistics and the residual statistics of each group of data based on the covariance matrix by using the abnormal monitoring algorithm based on multi-modal DPCA;

[0087] S240, judging whether an abnormal event will occur based on the Hotelling statistics and the residual statistics of the plurality of groups of data and the preset Hotelling statistics threshold and residual statistics threshold corresponding to the plurality of groups of data.

[0088] In one embodiment, the calculation method of the Hotelling statistics threshold and the residual statistics threshold of the plurality of groups of data comprises:

[0089] S241, obtaining a historical sample data set and dividing the historical sample data set into a plurality of initial data sets;

[0090] S242, calculating the covariance matrix of the plurality of initial data sets;

[0091] S243, calculating the eigenvalues and eigenvectors of the plurality of covariance matrices;

[0092] S244, sort the plurality of sets of characteristic values, screen the plurality of sets of characteristic values by using cumulative variance contribution rates, and merge the plurality of sets of initial data sets to generate a plurality of sets of intermediate data sets;

[0093] S245, model the plurality of sets of intermediate data sets to obtain a plurality of sets of PCA models, and calculate the residuals between the sets of PCA models;

[0094] S246, merge the plurality of sets of intermediate data sets based on a preset threshold and the residuals between the sets of PCA models to generate a new plurality of sets of data sets;

[0095] S247, repeat S242 to S246 until no new plurality of sets of data sets are generated, and then perform step S248;

[0096] S248, calculate the Hotelling statistic threshold and the residual statistic threshold of each set of PCA models by using the PCA method.

[0097] In one embodiment, in the offline detection stage, the training data is first initially grouped, then each set of data is merged according to the residual of each set of data, and each set of final data is modeled as a mode, and the maximum residual of each mode is saved as a threshold.

[0098] Specifically, in the initial grouping stage, as shown in Figure 2 The integrated historical data is set to a minimum grouping size H, and the massive data is automatically divided into groups, and the number of characteristic values of each group is calculated. If the contribution rate is greater than 99% (default value), the number of principal components is still equal to the number of input variables, and it is considered that the amount of data in the group is too small, and the group is merged with the next group. At the same time, the contribution rate and the number of principal components of the new merged group are calculated until the condition of an independent group is met. After the initial grouping is completed, the first group is modeled, the residual of the training set is calculated, and a threshold is automatically set according to the residual, such as the mean of the residual. The residual of the other group data input into the model established by the first group is calculated, and if the residual is less than the threshold, the first group is merged with the first group. The new first group is modeled again until no new group is generated. At this time, the first group data is modeled, the model is stored in the mode library, and the first group data is removed. The second group data is modeled, and the above operation is repeated until no new model is generated.

[0099] In order to better understand the flow of detection, the following will be combined Figure 3 to describe the fault detection process (offline modeling) in detail. The following is the establishment process of the fault monitoring model:

[0100] 1, the is expressed as an initial data group, wherein the superscript i represents the i-th grouping, and the subscript n idenotes the number of groups, and the PCA model is constructed by the following iterative process;

[0101] 2. For each group PCA model of obtained independently by the traditional PCA method, the number of retained principal components can be determined by cross-validation, and the test data set is taken as part of the training set;

[0102] 3. Similarity comparison is performed between two PCA models, and the residual error is taken as the similarity measure:

[0103]

[0104] 4. If the residual error is less than the threshold f, the corresponding data groups are merged, and the newly generated data groups are denoted as Go to step 2, otherwise stop.

[0105] 5. Determine the detection threshold: the threshold value is the maximum residual error of the corresponding modal training set itself.

[0106] On the basis of the above multi-modal PCA model, the multi-modal PCA-based anomaly detection method is as follows:

[0107] 1. When receiving online data, calculate the residual error SPE with each group modal by formula (1);

[0108] 2. Detection logic: if at least one residual error is less than the threshold, it is a normal sample, otherwise it is a fault sample.

[0109] In one embodiment, step S300 specifically comprises:

[0110] S310, when it is judged that an abnormal event will occur, the empirical mode decomposition method is used to denoise and trend extraction on the real-time data, to obtain the preprocessed data;

[0111] S320, input the preprocessed data into the pre-constructed linear model to obtain the estimated value of the target variable;

[0112] S330, based on the alarm threshold of the target variable and the estimated value of the target variable, the time of the abnormal event is calculated.

[0113] In one embodiment, the linear model is:

[0114] z i = φ + θt i = f(t i , φ, θ);

[0115] In the formula, φ, θ are random variables, and z iis the estimated value of the target variable at time i.

[0116] In one embodiment, the objective function of the linear model is:

[0117]

[0118] where y i is the true value of the target variable at time i, N is the window length, f(t i , φ, θ) is the estimated value of the target variable.

[0119] After detecting the abnormal working condition, the trend of the change of the key working condition parameter is predicted, and then the occurrence time of the automatic shutdown, pump shutdown and other events is predicted, which has important reference significance for timely maintenance after the abnormal working condition alarm. The difficulty of the abnormal working condition trend prediction lies in that the abnormal events are extremely rare in the historical data, and it is also difficult to verify whether the abnormal working condition data generated by the simulation machine simulation can accurately reflect the actual situation. Therefore, the present application realizes online adaptive modeling and prediction of the trend change of the key working condition parameter by using the online data after the abnormal working condition alarm under the condition that there is no abnormal working condition historical data. As shown in the following figure, Figure 4 the main steps include S310 trend extraction, S320 online modeling and S330 prediction of the remaining shutdown time (including the last two modules starting with prediction).

[0120] Since the data of the target parameter under the fault working condition and under the transition condition of the normal working condition is approximately linear, the present application adopts a linear model, so that the parameters solved are closer to the actual situation, and thus the prediction of the remaining shutdown time is more accurate.

[0121] Further, the noise and local oscillation contained in the real-time data will interfere with the trend prediction. Therefore, the present application adopts the empirical mode decomposition (EMD) method to denoise and extract the trend of the data.

[0122] Specifically, the Hilbert-Huang transform (HHT) based on EMD is a method for analyzing nonlinear and non-stationary signals. HHT contains two parts, namely EMD and Hilbert spectrum analysis (HAS). Its characteristics are that the nonlinear and non-stationary data are linearized and stabilized by EMD to obtain intrinsic mode function (IMF) components, and the characteristics of the data itself are preserved in the decomposition process. This decomposition method starts from the scale characteristics of the signal itself to decompose the signal, so that the characteristics of the signal can be analyzed as soon as possible, and has good local adaptability.

[0123] IMF extracted from the original signal by EMD method highlights the local characteristic information of the original signal. IMF has the following characteristics:

[0124] The number of zero-crossing points of the waveform is equal to the number of extreme points, or at most one difference, in the entire time history;

[0125] The average of the upper envelope line defined by the local maximum value and the lower envelope line defined by the local minimum value at any point on the signal is 0.

[0126] Among the above two limiting conditions, the first condition is actually very similar to the narrowband requirement of the commonly used stationary Gaussian process. The second condition is a new limiting condition, which changes the past global signal condition to a partial signal condition, which allows the signal to be decomposed and the correlation result to be obtained, and the decomposed multi-layer IMF is also a result with physical meaning.

[0127] However, not all signals meet the above-mentioned two IMF requirements, because a general signal at any time includes more than one mode.

[0128] The EMD method decomposes the signal based on the following three assumptions:

[0129] The signal has at least one maximum point and one minimum point;

[0130] The characteristic time scale is set to the time interval between adjacent extreme points;

[0131] If the signal encounters a situation without extreme points but only inflection points, the original signal is first differentiated once or multiple times to obtain extreme points, and then the integral of the resulting result can obtain the corresponding component.

[0132] EMD is specifically decomposed by a process called "sifting". The flow chart is shown in Figure 5 , and the specific operation process is as follows:

[0133] The first step of the process is to find all the maximum points and minimum points of the original signal s(t), and then connect all the maximum points with a cubic spline curve to form the upper envelope line of s(t), and connect all the minimum points with the same method to obtain the lower envelope line of s(t). According to the local maximum and local minimum of the signal s(t), the average value of the upper envelope v1(t) and the lower envelope v2(t) is calculated:

[0134]

[0135] Then find the difference h 11 between s(t) and m 11 :

[0136] h 11 = s(t) - m 11 (Equation 3)

[0137] In the ideal case, m 11 should satisfy the definition of IMF, because h 11 is constructed to satisfy the symmetry. After the first iteration, this value has a probability of becoming a local maximum. The new local maximum point produced in Equation (3) indicates that the required correct mode is missing in the initial step. Through repeated decomposition and extraction, the extraction process can recover the original signal and express it as a low-amplitude carrier wave.

[0138] The extraction process has two purposes: to eliminate the carrier wave and to make the waveform more symmetrical. Around these two purposes, the extraction process must be repeated the required number of times so that the subtraction of the extracted signal becomes an IMF. In the subsequent extraction process, h 11 can only be considered as an IMF prototype if h 11 is not an IMF, it is regarded as a new s(t) and Equation (2) is repeated 2k times.

[0139] h 1k = h 1(k-1) - m 1k (Equation 4)

[0140] h 1k and h 1(k-1) in Equation (4) are the data obtained from the kth and (k-1)th screening, respectively, and m 1k is the midline (average) of the upper and lower envelopes of h 1(k-1) . The S D value calculated by Equation (5) can be used to determine whether the result of each "screening" is a required IMF component:

[0141]

[0142] A key stopping criterion is required here: if the S D value is between 0.2 and 0.3, let c1 = h 1k be considered as an IMF, make r = s(t) - c1, and then regard r as a new s(t) to repeat the above process to obtain the second IMF c2 , the third IMF c3 … until r(t) becomes a monotonic function or the absolute value of r(t) is less than a specific threshold value that can be considered as a measurement error, at which point the process stops. Thus, the signal obtained by superimposing each layer of IMF components and the residue is which shows the completeness of EMD decomposition. In the above equation, r is the residual, representing the average trend in the signal, and each IMF component c1 represents the content waveform of each frequency band signal, respectively. The smaller the value of i, the higher the frequency. The distribution of different frequency elements at different times varies with the signal itself.

[0143] After EMD processing, the component representing the overall trend in IMF is extracted as the preprocessed data, which is input into the subsequent predictor.

[0144] When the preprocessed data is obtained, the target variable trend curve at time t i can be represented by the following linear model:

[0145] z i = f(t i , φ, θ) i , φ, θ) (Equation 6)

[0146] In the equation, φ and θ are random variables, and z i is the estimated value of the target variable at time i. By introducing a time domain window, the objective function of the linear model can be described as an online optimization problem:

[0147]

[0148] where y i is the true value of the target variable at time i, N is the window length, and f(t i , φ, θ) is the estimated value of the target variable. The optimal solution at time k is denoted as (φ * , θ * ) k , which is substituted into the linear model to obtain the estimated value z k of the target variable at time k.

[0149] In the process of online prediction, the above linear model should be constantly updated based on the data in the past period of time. Therefore, the key working condition parameter trend prediction algorithm based on the adaptive exponential model is as follows (the algorithm used in step S300):

[0150] 1. Collect data z k at time t k with a window length of s, and perform empirical mode decomposition to extract the IMF component containing the overall trend.

[0151] 2. Assign initial values to the linear model parameters.

[0152] 3. At time k, based on the data ks from time t k to t i , train a linear model f(t i , φ, θ), φ, θ).

[0153] 4. The linear model f(t i , φ, θ) can estimate the prediction value of k s th moment That is:

[0154]

[0155] 5. Wait for the next moment data, jump to step 1 (the data collection part above).

[0156] At k moment, solve formula (7) to get the optimal solution (φ * , θ * ) k . The estimated value of the target variable at k moment can be obtained:

[0157] z k = φ * + θ * t k ;

[0158] Given the alarm threshold ξ of the target variable, it satisfies:

[0159] ξ = φ * + θ * (t k + T k );

[0160] The remaining shutdown time can be obtained:

[0161]

[0162] In the formula, T k represents the time of abnormal event, ξ represents the alarm threshold of the target variable, φ * , θ * are the optimal solutions of random variables at k moment.

[0163] The length of the time window dynamically changes in the prediction process, and its calculation logic is to obtain the latest trend characteristics of the data as much as possible. Therefore, the algorithm requires the absolute value of the difference between the first and last values of the time window to be less than a threshold, which can ensure that the data trend can be captured as soon as possible after it changes.

[0164] The application establishes an early warning and diagnosis model by monitoring the operation state of the power plant. By means of data-driven process monitoring, high-dimensional data is transformed into low-dimensional data, and information is obtained therefrom to capture the abnormality of the correlation between process variables, thereby informing the operator to respond in time. The abnormality of the operation process is monitored, analyzed and warned, and problems including whether the current operation state is normal and the time of the abnormal event are predicted, thereby providing auxiliary support for ensuring the stable and safe operation of the power plant.

[0165] In addition, based on the above-mentioned event warning method based on nuclear power control logic, the application also correspondingly provides an event warning device based on nuclear power control logic, as shown in the accompanying drawings, comprising: Figure 6

[0166] A data acquisition module 11 is configured to acquire real-time data.

[0167] An abnormality judgment module 12 is configured to calculate the real-time data by using an abnormality monitoring algorithm based on multi-modal DPCA to determine whether an abnormal event will occur, wherein the abnormal event is at least one of a shutdown event, a shutdown event and a load rejection event.

[0168] An abnormality prediction module 13 is configured to predict the time of occurrence of the abnormal event when it is determined that an abnormal event will occur by using a key working condition parameter trend prediction algorithm based on an adaptive linear model to predict the trend of change of the real-time data.

[0169] An output module 14 is configured to output and alarm the time of occurrence of the abnormal event.

[0170] In one embodiment, the abnormality judgment module 12 is specifically configured to:

[0171] group the real-time data to obtain a plurality of groups of data;

[0172] calculate the covariance matrix of the plurality of groups of data by using the abnormality monitoring algorithm based on multi-modal DPCA;

[0173] calculate the Hotelling statistics and the residual statistics of each group of data based on the covariance matrix by using the abnormality monitoring algorithm based on multi-modal DPCA;

[0174] determine whether an abnormal event will occur based on the Hotelling statistics and the residual statistics of the plurality of groups of data and the preset Hotelling statistics threshold and the residual statistics threshold corresponding to the plurality of groups of data.

[0175] In one embodiment, the calculation method of the Hotelling statistics threshold and the residual statistics threshold of the plurality of groups of data comprises:

[0176] ​S241, acquire a historical sample data set, divide the historical sample data set into a plurality of initial data sets;

[0177] S242, calculate a plurality of covariance matrices of the initial data sets;

[0178] S243, calculate a plurality of eigenvalues and eigenvectors of the plurality of covariance matrices;

[0179] S244, sort the plurality of eigenvalues, filter the plurality of eigenvalues by using a cumulative variance contribution rate, and generate a plurality of intermediate data sets after merging the plurality of initial data sets;

[0180] S245, model the plurality of intermediate data sets, calculate a plurality of PCA models, and calculate a plurality of residuals between the plurality of PCA models;

[0181] S246, merge the plurality of intermediate data sets based on a preset threshold and the plurality of residuals between the plurality of PCA models, and generate a plurality of new data sets;

[0182] S247, repeat S242 to S246 until no new plurality of data sets is generated, and execute step S248;

[0183] S248, calculate Hotelling statistics and residual statistics of the plurality of PCA models by using a PCA method.

[0184] In one embodiment, the anomaly prediction module 13 is specifically configured to:

[0185] When it is determined that an abnormal event will occur, an empirical mode decomposition method is used to denoise and extract a trend from real-time data to obtain preprocessed data;

[0186] The preprocessed data is input into a pre-constructed linear model to obtain an estimated value of a target variable;

[0187] Based on an alarm threshold of the target variable and the estimated value of the target variable, a time of occurrence of the abnormal event is calculated.

[0188] In one embodiment, the linear model is:

[0189] z i = f (t i , φ, θ) i

[0190] In the formula, φ and θ are random variables, z i is an estimated value of the target variable at time i.

[0191] In one embodiment, the objective function of the linear model is: ​

[0192]

[0193] In the formula, y i is the true value of the target variable at time i, N is the window length, f(t i , φ, θ) is the estimated value of the target variable.

[0194] In one embodiment, the formula for calculating the time of the abnormal event is:

[0195]

[0196] In the formula, T k represents the time of the abnormal event, ξ represents the alarm threshold of the target variable, φ * , θ * is the optimal solution of the random variable at time k.

[0197] In summary, the event early warning method and device based on nuclear power control logic provided by the application first acquires real-time data; then uses an abnormal monitoring algorithm based on multi-modal DPCA to calculate the real-time data to determine whether an abnormal event will occur; then when it is determined that an abnormal event will occur, a key working condition parameter trend prediction algorithm based on an adaptive linear model is used to predict the change trend of the real-time data to predict the time of the abnormal event; finally, the time of the abnormal event is output and an alarm is given. By performing abnormal detection on event-related parameters, input parameter ranges are divided according to different events, data processing, feature extraction, multivariate statistical analysis, dynamic principal component analysis, and other algorithms are used to carry out research on shutdown events and four-scenario abnormal detection, then trend prediction is introduced to determine how long the abnormal event will occur, which is based on historical power plant operation data and parameter correlation to perform regression modeling, predicts the future operation trend of the related parameters of a specific event within a period of time, and finally merges the information into early warning information, thereby providing auxiliary support for ensuring the smooth and safe operation of the power plant.

[0198] In addition, the application also provides a computer device, which comprises a memory and a processor, the memory stores computer readable instructions, and the processor implements the steps of the event early warning method based on nuclear power control logic when executing the computer readable instructions.

[0199] In addition, the application also provides a computer readable storage medium, which stores computer readable instructions, and the computer readable instructions implement the steps of the event early warning method based on nuclear power control logic when executed.

[0200] Computer-readable media includes permanent and non-permanent, movable and non-movable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information accessible by a computing device.

[0201] The above description is merely a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or replacements within the technical scope disclosed by the present application can be easily thought of by those skilled in the art, and should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. An event early warning method based on nuclear power control logic, characterized in that: include: S100, obtaining real-time data; S200: Calculate the real-time data to determine whether an abnormal event occurs, including: S210, grouping the real-time data to obtain several groups of data; S220, calculating the covariance matrix of several groups of data using an anomaly monitoring algorithm based on multimodal DPCA; S230, based on the covariance matrix, using the anomaly monitoring algorithm based on multimodal DPCA to calculate the Hotelling statistic and the residual statistic of each group of data; S240, judging whether an abnormal event will occur based on the Hotelling statistics and residual statistics of the plurality of data sets and the preset Hotelling statistics thresholds and residual statistics thresholds corresponding to the plurality of data sets; S300, when it is determined that an abnormal event will occur, using a key operating condition parameter trend prediction algorithm based on an adaptive linear model to predict the change trend of real-time data to predict the time when the abnormal event will occur; S400: Output the time when the abnormal event occurs and generate an alarm.

2. The event warning method based on nuclear power control logic according to claim 1 is characterized in that: In S200, an abnormality monitoring algorithm based on multi-modal DPCA is used to monitor abnormal events during the operation of the unit, where the abnormal events include at least one of a reactor shutdown event, a shutdown event, and a load rejection event.

3. The event warning method based on nuclear power control logic according to claim 1 is characterized in that: The S240 includes: S241, obtaining a historical sample data set, and dividing the historical sample data set into several groups of initial data sets; S242, calculating the covariance matrix of several groups of initial data sets; S243, calculating the eigenvalues ​​and eigenvectors of several groups of covariance matrices; S244, sorting the plurality of sets of eigenvalues, screening the plurality of sets of eigenvalues ​​using cumulative variance contribution rates, merging the plurality of sets of initial data sets, and generating a plurality of sets of intermediate data sets; S245, modeling the multiple sets of intermediate data sets to obtain multiple sets of PCA models, and then calculating the residuals between the PCA models; S246, merging the multiple intermediate data sets based on a preset threshold and the residuals between the PCA models to generate multiple new data sets; S247, repeat S242 to S246 until no new multiple data sets are generated, then execute S248; S248. Using the PCA method, the Hotelling statistic threshold and the residual statistic threshold of each group of PCA models are finally calculated.

4. The event warning method based on nuclear power control logic according to claim 1 is characterized in that: The process of establishing a fault detection model includes: (1) Denoted as the initial data set, the PCA model is constructed through the following iterative process; (2) For each group PCA model The number of principal components retained was determined by cross-validation and the test dataset was used as part of; (3) Compare the similarity between the two PCA models, using the residual as the similarity measure: (4) If the residual is less than the threshold f, the corresponding data groups are merged and these newly generated data groups are expressed as Go to step (2), otherwise stop; (5) Determine the detection threshold: The value of the threshold is the maximum residual of the corresponding modality training set itself.

5. The event warning method based on nuclear power control logic according to claim 1 is characterized in that: S300 includes: S310: When it is determined that an abnormal event will occur, the empirical mode decomposition method is used to remove noise and extract trends from the real-time data to obtain preprocessed data; S320, inputting the preprocessed data into a pre-built linear model to obtain an estimated value of the target variable; S330: Calculate the time when the abnormal event occurs based on the alarm threshold of the target variable and the estimated value of the target variable.

6. The event warning method based on nuclear power control logic according to claim 1 is characterized in that: In S300, the key operating parameter trend prediction algorithm based on the adaptive exponential model includes: (1) Collect t k At time t, the data z with window length s k , perform empirical mode decomposition on it and extract the IMF component containing the overall trend; (2) Assign initial values ​​to the linear model parameters; (3) At time k, based on t ks to t k Data at all times Train a linear model f(t i ,φ,θ); (4) Through the linear model f(t i ,φ,θ) estimate k s The predicted value at time k as follows: (5) Wait for the next moment data to arrive and jump to step (1); At time k, the pair Solve and get the optimal solution (φ * ,θ * ) k , get the estimated value of the target variable at time k: z k =φ * +θ * t k ; The alarm threshold ξ of the known target variable satisfies: ξ=φ * +θ * (t k +T k ); Calculate the remaining shutdown time: Where, T k represents the time when the abnormal event occurs, ξ represents the alarm threshold of the target variable, φ * ,θ * represents the optimal solution of the random variable at time k.

7. An event warning device based on nuclear power control logic, characterized in that: The event warning method based on nuclear power control logic according to any one of claims 1 to 6 comprises: Data acquisition module, used to obtain real-time data; The anomaly judgment module is used to calculate the real-time data using the anomaly monitoring algorithm based on multimodal DPCA to determine whether an abnormal event will occur; The abnormality prediction module is used to predict the changing trend of real-time data and the time when the abnormal event will occur by using the key operating parameter trend prediction algorithm based on the adaptive linear model when it is judged that an abnormal event will occur; The output module is used to output the time when the abnormal event occurs and issue an alarm.

8. A computer device comprising a memory and a processor, wherein the memory stores computer-readable instructions, wherein: When the processor executes the computer-readable instructions, the steps of the event warning method based on nuclear power control logic described in any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having computer-readable instructions stored thereon, characterized in that: When the computer-readable instructions are executed, the steps of the event warning method based on nuclear power control logic described in any one of claims 1-6 are implemented.

Images