Method and system for identifying evolution law of multi-monitoring data based on waveform alignment

Abstract

The present invention discloses a method and system for identifying the evolution law of multi-monitoring data based on waveform alignment. The method includes: acquiring monitoring data and performing abnormal data processing on the monitoring data; performing normalization processing on the data after the abnormal data processing is completed; performing data curve fitting on the data after the normalization processing is completed to obtain a fitting curve; collecting the fitting curve, performing correlation analysis on the pairwise combinations of measurement points in the fitting curve, and if the obtained correlation coefficient is higher than a preset threshold, calculating the displacement between the waveforms of the pairwise measurement points, and calculating the delay between the pairwise measurement points through the displacement. By implementing the present invention, the influence caused by abnormal data in the measurement point data is avoided, the correlation degree of the measurement points is preliminarily screened based on the correlation threshold, and then the curve alignment calculation is performed, which greatly improves the calculation efficiency of waveform alignment of the measurement point data.

Description

Technical Field

[0001] The present invention relates to the technical field of computer software development, and in particular to a method and system for identifying the evolution law of multi-monitoring data based on waveform alignment. Background Art

[0002] In a nuclear power plant, due to the large number of process systems and complex equipment in the nuclear power plant, during the operation of the equipment, the equipment often stops running due to system or equipment failures. After the equipment fails, it takes a lot of time for professional personnel such as operation and maintenance to check the cause of the failure, and the cause of the failure cannot be quickly diagnosed, which seriously affects the operation and recovery of the equipment. If the equipment status can be monitored, the failure omen can be detected in time, and the occurred failure can be diagnosed, it will greatly improve the reliability and stability of the equipment operation and avoid greater economic losses.

[0003] For equipment status monitoring and fault diagnosis, it is necessary to judge the internal working status of the equipment or the loss status of the mechanical structure according to the change of the equipment measuring point data, determine the nature, degree, category and location of the fault, predict its development trend, and study the mechanism of fault generation. Therefore, a method is needed to analyze the change of the equipment measuring point data to realize equipment status monitoring and fault diagnosis. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method and system for identifying the evolution law of multi-monitoring data based on waveform alignment for at least one defect existing in the related technologies mentioned in the above background art: analyzing the change of the equipment measuring point data to realize equipment status monitoring and fault diagnosis.

[0005] The technical solution adopted by the present invention to solve its technical problems is: providing a method for identifying the evolution law of multi-monitoring data based on waveform alignment, including the following steps:

[0006] S10: Obtain the monitoring data of the nuclear power plant equipment, and perform abnormal data processing on the monitoring data;

[0007] S20: Normalize the data after the abnormal data processing is completed;

[0008] S30: Perform data curve fitting on the data after the normalization processing is completed to obtain a fitting curve;

[0009] S40: Collect the fitting curve, perform correlation analysis on the pairwise measurement point combinations in the fitting curve. If the obtained correlation coefficient is higher than a preset threshold, calculate the displacement between the waveforms of the pairwise measurement points, and calculate the delay between the pairwise measurement points through the displacement.

[0010] Preferably, in the method for identifying the evolution law of multi-monitoring data based on waveform alignment according to the present invention, step S10 further includes:

[0011] S101: Obtain the time series data sequence of each measuring point in the monitoring data, set the window range, perform sliding calculation on the original data of all the measuring points according to the window range, and calculate the average value and standard deviation of the data within the window range;

[0012] S102: If the data of the measuring point at the center of the window range is not within the range set based on the average value and the standard deviation, then determine that the data of the measuring point is abnormal data;

[0013] S103: Process the abnormal data by using the median filtering algorithm.

[0014] Preferably, in the method for identifying the evolution law of multi-monitoring data based on waveform alignment according to the present invention, step S30 includes:

[0015] Set a polynomial function, obtain the optimal parameter solution of the polynomial function by using the least squares method, and further obtain the fitting curve.

[0016] Preferably, in the method for identifying the evolution law of multi-monitoring data based on waveform alignment according to the present invention, the case where the correlation coefficient obtained by analysis is higher than the preset threshold includes:

[0017] If the absolute value of the correlation coefficient obtained by analysis is higher than the preset threshold;

[0018] Wherein, if the correlation coefficient is positive, it is determined that the change of the two measuring points is positively correlated; if the correlation coefficient is negative, it is determined that the change of the two measuring points is negatively correlated.

[0019] Preferably, in the method for identifying the evolution law of multi-monitoring data based on waveform alignment according to the present invention, the calculating the delay between the two measuring points through the displacement includes:

[0020] Based on the calculation results of the delays between all the two measuring points, group the measuring points based on the correlation coefficient, divide the measuring points into relevant combinations, calculate the reference measuring point in the combination, and recalculate the delay of the measuring points according to the reference measuring point.

[0021] The present invention also constructs a system for identifying the evolution law of multi-monitoring data based on waveform alignment, including:

[0022] An abnormal data processing module, configured to obtain the monitoring data of nuclear power plant equipment and perform abnormal data processing on the monitoring data;

[0023] A normalization processing module for normalizing the data after abnormal data processing;

[0024] A curve fitting module for performing data curve fitting on the data after normalization processing to obtain a fitting curve;

[0025] A waveform alignment module for collecting the fitting curve, performing correlation analysis on pairwise measurement point combinations in the fitting curve, and if the obtained correlation coefficient is higher than a preset threshold, calculating the displacement between the waveforms of the pairwise measurement points, and calculating the delay between the pairwise measurement points through the displacement.

[0026] Preferably, in the multi-monitoring data evolution law recognition system based on waveform alignment of the present invention, the abnormal data processing module further includes:

[0027] A calculation unit for obtaining the time series data sequence of each measurement point in the monitoring data, setting a window range, performing sliding calculation on all the original measurement point data according to the window range, and calculating the average value and standard deviation of the data within the window range;

[0028] A judgment unit for determining that the measurement point data is abnormal data if the measurement point data at the center of the window range is not within the range set based on the average value and the standard deviation;

[0029] A processing unit for processing the abnormal data by using a median filtering algorithm.

[0030] Preferably, in the multi-monitoring data evolution law recognition system based on waveform alignment of the present invention, the curve fitting module includes:

[0031] Setting a polynomial function, obtaining the optimal parameter solution of the polynomial function by using the least squares method, and further obtaining the fitting curve.

[0032] Preferably, in the multi-monitoring data evolution law recognition system based on waveform alignment of the present invention, the case where the obtained correlation coefficient is higher than a preset threshold includes:

[0033] Whether the absolute value of the obtained correlation coefficient is higher than the preset threshold;

[0034] Wherein, if the correlation coefficient is positive, it is determined that the change of the pairwise measurement points is positively correlated; if the correlation coefficient is negative, it is determined that the change of the pairwise measurement points is negatively correlated.

[0035] Preferably, in the multi-monitoring data evolution law recognition system based on waveform alignment of the present invention, the calculating the delay between the pairwise measurement points through the displacement includes:

[0036] Based on the delay calculation results between all pairs of measurement points, group the measurement points based on the correlation coefficient, divide the measurement points into correlated combinations, calculate the reference measurement points in the combinations, and recalculate the delays of the measurement points based on the reference measurement points.

[0037] By implementing the present invention, the following beneficial effects are achieved:

[0038] The present invention discloses a method and system for identifying the evolution law of multi-monitoring data based on waveform alignment. By performing abnormal data processing and normalization processing on the discrete data points of the monitoring data, curve fitting is carried out to obtain a fitting curve. Correlation analysis is performed on two measurement points within the fitting curve. When the correlation coefficient is higher than a preset threshold, it is considered that the correlation degree is high, and then the displacement between the measurement point waveforms is calculated, thereby measuring the correlation and delay between the two measurement points. By implementing the present invention, the correlation degree of the measurement points is preliminarily screened based on the correlation threshold, and then curve alignment calculation is performed, greatly improving the calculation efficiency of waveform alignment of the measurement point data. Moreover, the identification method is encapsulated into a general method, solving the problem that it cannot be directly applied due to complex business logic and involving multi-layer data transformation processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The following will further illustrate the present invention in conjunction with the drawings. In the drawings:

[0040] Figure 1 is a flowchart of the method for identifying the evolution law of multi-monitoring data based on waveform alignment of the present invention;

[0041] Figure 2 is a normal distribution probability diagram of the present invention;

[0042] Figure 3 is a curve fitting schematic diagram of the present invention;

[0043] Figure 4 is a flowchart of the waveform alignment calculation of the data of the present invention;

[0044] Figure 5 is a schematic diagram of the measurement point delay analysis of the present invention;

[0045] Figure 6 is a front-end display diagram of the measurement point analysis system of the present invention;

[0046] Figure 7 is a block diagram of the modules of the system for identifying the evolution law of multi-monitoring data based on waveform alignment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0047] In order to have a clearer understanding of the technical features, objectives, and effects of the present invention, the specific embodiments of the present invention will now be described in detail with reference to the drawings.

[0048] It should be noted that the flowcharts shown in the accompanying drawings are only exemplary and do not necessarily include all the contents and operations / steps, nor do they have to be executed in the order described. For example, some operations / steps can be decomposed, while some operations / steps can be combined or partially combined, so the actual execution order may change according to actual conditions.

[0049] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities may be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0050] In this embodiment, if Figure 1 As shown, the present invention provides a method for identifying evolution rules of multiple monitoring data based on waveform alignment, comprising the following steps:

[0051] S10: Acquire monitoring data of nuclear power plant equipment and perform abnormal data processing on the monitoring data;

[0052] S20: normalizing the data after abnormal data processing;

[0053] S30: performing data curve fitting on the normalized data to obtain a fitting curve;

[0054] S40: collecting the fitting curve, performing correlation analysis on the combination of two measuring points in the fitting curve, and if the correlation coefficient obtained by the analysis is higher than a preset threshold, calculating the displacement between the waveforms of the two measuring points, and calculating the delay between the two measuring points through the displacement.

[0055] Specifically:

[0056] According to the excellent properties of normal distribution, The criterion is often used to determine whether the data is abnormal. Symmetrical, such as Figure 2 As shown, the values are distributed in The probability of is 0.6827, and the value is distributed in The probability of being in the mean is 0.9973. That is, only 0.3% of the data will fall within the mean. In addition, this is a low-probability event. In order to prevent extreme values from affecting the overall robustness of the model, they are often identified as outliers and removed from the data.

[0057] In this embodiment, step S10 further includes:

[0058] S101: Obtain the time series data sequence of each measurement point in the monitoring data of nuclear power plant equipment, set the window range, perform sliding calculations on the original data of all measurement points according to the window range, and calculate the average value of the data within the window range. and standard deviation ;

[0059] S102: If the measurement point data at the center of the window range is not within the range set based on the average value and standard deviation, and its range is expressed as , then determine that the measurement point data is abnormal data;

[0060] S103: Process the abnormal data using the median filtering algorithm, that is, within a sliding window centered on the abnormal data, sort all the data, and replace the abnormal data with the median value of the window data to complete the smoothing process of the abnormal data; for non-abnormal data, no processing is performed.

[0061] The data of different measurement points often have different dimensions and dimension units, which will affect the results of data analysis. In order to eliminate the influence of dimensions between measurement points, data standardization processing is required to make each measurement point at the same order of magnitude for comprehensive comparison and evaluation. The method of normalization processing is used to solve this problem. The purpose of normalization is to limit the preprocessed data within a certain range (such as [0,1] or [-1,1]), so as to reduce the influence caused by the large difference in the data values of different measurement points.

[0062] The normal operation state data includes the original data of multiple measurement points. The following formula is used to perform normalization processing on the normal operation state data:

[0063]

[0064] where is the measurement point serial number, is the data after normalization processing of the th measurement point, is the original data of the th measurement point, max( ) is the maximum value, min( ) is the minimum value.

[0065] Data curve fitting is a data processing method that uses a continuous curve to approximately depict or simulate the functional relationship between the coordinates represented by a discrete point group on a plane. Measuring points of a device often can only obtain several discrete data through methods such as sampling. Based on these data, if a continuous function or a more dense discrete equation can be found such that the measuring point data and the curve of the equation can approximately coincide to the greatest extent, then mathematical calculations can be performed on the data according to the curve equation, and calculation and analysis can be carried out on the data results.

[0066] Moreover, in this embodiment, step S30 includes:

[0067] Set a polynomial function, obtain the optimal parameter solution of the polynomial function by using the least squares method, and then obtain the fitting curve. Let the curve function be an n-degree polynomial:

[0068]

[0069] where is the polynomial function, and a set of parameter solutions is obtained by using the least squares method such that is the minimum.

[0070] After calculating the parameter solutions, the fitting curve is obtained, and the visualization result is as shown in Figure 3 . The discrete points are located within the fitting curve.

[0071] In addition, in this embodiment, if the correlation coefficient obtained by analysis is higher than a preset threshold, it includes:

[0072] If the absolute value of the correlation coefficient obtained by analysis is higher than the preset threshold;

[0073] Among them, if the correlation coefficient is positive, it is determined that the change between two measuring points is positively correlated; if the correlation coefficient is negative, it is determined that the change between two measuring points is negatively correlated.

[0074] Furthermore, the delay between two measuring points is calculated through displacement, including:

[0075] Based on the calculation results of the delays between all pairs of measuring points, grouping of the measuring points is performed based on the correlation coefficient, the measuring points are divided into correlated combinations, and the reference measuring point in the combination is calculated. The delay of the measuring points is recalculated based on the reference measuring point.

[0076] The specific steps for its waveform alignment are as shown in Figure 4 :

[0077] Sample multiple fitting curves formed by multiple measurement points. Set the sampling frequency to the maximum frequency in the original measurement point data to obtain time-series data with consistent time points and frequencies for multiple measurement points, that is, adjust the time-series data of all measurement points to the same frequency and time point, ensuring the comparability of the time-series data. Otherwise, data at different time points cannot be compared;

[0078] Combine the measurement points in pairs for calculation and calculate the correlation coefficient between each pair based on the correlation algorithm. Preferably, use the Pearson similarity algorithm to calculate the correlation coefficient, and no detailed analysis will be done here;

[0079] Preferably, when the absolute value of the correlation coefficient exceeds 0.85, it is considered that the measurement point combination has a high correlation and the next calculation can be carried out;

[0080] After obtaining the measurement point combination [a, b] with a correlation coefficient exceeding 0.85 in the previous step, the delay in the data changes of measurement points a and b will be further calculated. If there is a delay between measurement points a and b, then at the same time point, the correlation coefficient of the data curves of a and b will not be the maximum value; after shifting the data curve of b in time by the delay duration, the resulting time-series curve b` and the time-series curve a will have a maximum correlation coefficient. Therefore, the delay calculation method is as follows: taking measurement point a as the reference, divide [-100, 100] into a total of 201 time intervals, shift measurement point b according to the time intervals, and then calculate the correlation degree. If the obtained correlation coefficient > is obtained, then keep this value and calculate successively to obtain and is the delay time between measurement point a and measurement point b.

[0081] After obtaining the calculation results of each pair of measurement point combinations, group the measurement points based on the correlation coefficient results, divide the measurement points into relevant combinations, and calculate the reference measurement point in the combination. Recalculate the delay of each measurement point based on the reference measurement point. The delay is expressed as follows Figure 5 indicated.

[0082] Such as Figure 6 shown, it is the result graph displayed at the front end of the system. By calculating the delay, it is analyzed whether the changes of the measurement points are relevant, the order before and after the changes of the measurement points, the time interval of the changes of the measurement points, etc. These characteristic information of the changes of the measurement points are exactly the data basis for realizing equipment status monitoring and fault diagnosis.

[0083] In this embodiment, as Figure 7 shown, the present invention also constructs a multi-monitoring data evolution law recognition system based on waveform alignment, including:

[0084] An abnormal data processing module, which is used to obtain the monitoring data of nuclear power plant equipment and perform abnormal data processing on the monitoring data;

[0085] A normalization processing module for normalizing the data after abnormal data processing;

[0086] A curve fitting module for performing data curve fitting on the data after normalization processing to obtain a fitting curve;

[0087] A waveform alignment module for collecting the fitting curve, performing correlation analysis on pairwise measurement point combinations in the fitting curve. If the obtained correlation coefficient is higher than a preset threshold, calculate the displacement between the waveforms of pairwise measurement points, and calculate the delay between pairwise measurement points through the displacement.

[0088] Specifically:

[0089] According to the excellent properties of the normal distribution, the 3σ criterion is often used to determine whether data is abnormal. Since the normal distribution is symmetric about the mean μ, as Figure 2 shown, the probability that the numerical distribution is in (μ - σ, μ + σ) is 0.6827, and the probability that the numerical distribution is in (μ - 3σ, μ + 3σ) is 0.9973. That is, only 0.3% of the data will fall outside ±3σ of the mean, which is a small probability event. To avoid the influence of extreme values on the overall robustness of the model, they are often determined as outliers and removed from the data.

[0090] In this embodiment, the abnormal data processing module further includes:

[0091] A calculation unit for obtaining the time series data sequence of each measurement point in the monitoring data of nuclear power plant equipment, setting a window range, performing sliding calculation on all measurement point raw data according to the window range, and calculating the average value of the data within the window range and the standard deviation ;

[0092] A judgment unit for determining that the measurement point data is abnormal data if the measurement point data at the center of the window range is not within the range set based on the average value and the standard deviation, and the range is expressed as (μ - 3σ, μ + 3σ);

[0093] A processing unit for processing abnormal data using a median filtering algorithm, that is, sorting all data within a sliding window centered on the abnormal data, and replacing the abnormal data with the median value of the window data to complete the smoothing processing of the abnormal data; for non-abnormal data, no processing is performed.

[0094] The data of different measurement points often have different dimensions and dimension units, which will affect the results of data analysis. In order to eliminate the dimension influence between measurement points, data standardization processing is required to bring each measurement point to the same order of magnitude for comprehensive comparison and evaluation, and the normalization method is used to solve this problem. The purpose of normalization is to limit the preprocessed data within a certain range (such as [0, 1] or [-1, 1]), so as to reduce the influence caused by the large difference in the data values of different measurement points.

[0095] The normal operation state data includes the original data of multiple measurement points, and the following formula is used to normalize the normal operation state data:

[0096]

[0097] where is the measurement point serial number, is the data after normalization processing of the th measurement point, is the original data of the th measurement point, max( ) is the maximum value, and min( ) is the minimum value.

[0098] Data curve fitting is a data processing method that uses a continuous curve to approximately depict or compare the functional relationship between the coordinates represented by a discrete point group on a plane. The measurement points of a device can often only obtain a number of discrete data through methods such as sampling. Based on these data, if a continuous function or a more dense discrete equation can be found such that the measurement point data and the curve of the equation can be approximately coincident to the greatest extent, then mathematical calculations can be performed on the data according to the curve equation, and the data results can be calculated and analyzed.

[0099] Moreover, in this embodiment, the curve fitting module includes:

[0100] Set a polynomial function, and use the least squares method to obtain the optimal parameter solution of the polynomial function, and then obtain the fitting curve. Let the curve function be an nth-degree polynomial:

[0101]

[0102] where is the polynomial function, and a set of parameter solutions are obtained using the least squares method such that is the minimum.

[0103] After calculating the parameter solutions, the fitting curve is obtained, and the visualization result is as shown in Figure 3 , and the discrete points are located within the fitting curve.

[0104] In addition, in this embodiment, if the correlation coefficient obtained by analysis is higher than the preset threshold, it includes:

[0105] Whether the absolute value of the correlation coefficient obtained by analysis is higher than the preset threshold;

[0106] Among them, if the correlation coefficient is positive, it is determined that the changes of two measurement points are positively correlated; if the correlation coefficient is negative, it is determined that the changes of two measurement points are negatively correlated.

[0107] Furthermore, the delay between two measurement points is calculated through displacement, including:

[0108] Based on the calculation results of the delays between all pairs of measurement points, the measurement points are grouped according to the correlation coefficient, the measurement points are divided into correlated combinations, and the reference measurement points in the combination are calculated, and the delays of the measurement points are recalculated based on the reference measurement points.

[0109] The specific steps for its waveform alignment are as Figure 4 shown:

[0110] Sample the multiple fitting curves formed by multiple measurement points, and set the sampling frequency as the maximum frequency in the original measurement point data, and obtain the time-series data with consistent time points and frequencies for multiple measurement points, that is, adjust the time-series data of all measurement points to the same frequency and time point, ensuring the comparability of the time-series data. Otherwise, data at different time points cannot be compared;

[0111] Combine the measurement points in pairs for calculation, and calculate the correlation coefficient between each pair based on the correlation algorithm; preferably, use the Pearson similarity algorithm to calculate the correlation coefficient, and no detailed analysis is made here;

[0112] Preferably, when the absolute value of the correlation coefficient exceeds 0.85, it is considered that the measurement point combination has a high degree of correlation and the next calculation can be performed;

[0113] After obtaining the measurement point combination [a, b] with a correlation coefficient exceeding 0.85 in the previous step, the delay of the data changes of measurement points a and b will be further calculated. If there is a delay between measurement points a and b, then at the same time point, the correlation coefficient of the data curves of a and b will not be the maximum value; after translating the data curve of b in time by the delay duration, the resulting time-series curve b' and the time-series curve a will have a maximum correlation coefficient. Therefore, the delay calculation method is as follows: taking measurement point a as the reference, divide [-100, 100] into a total of 201 time intervals, translate measurement point b according to the time intervals, and then perform the correlation degree calculation. If the obtained correlation coefficient Pn > P0, then retain this value, calculate sequentially to obtain max(Pn), and n is the delay time between measurement point a and measurement point b.

[0114] After obtaining the calculation results of pairwise measurement point combinations, group the measurement points based on the correlation coefficient results, divide the measurement points into correlated combinations, and calculate the reference measurement points in the combinations. Recalculate the delays of each measurement point with the reference measurement points. The delay is expressed as follows Figure 5 It is shown as

[0115] As Figure 6 shown, it is the result graph displayed by the front end of the system. By calculating the delay, analyze whether the changes of the measurement points are relevant, the order before and after the change of the measurement points, the time interval of the change of the measurement points, etc. These characteristic information of the changes of the measurement points are exactly the data basis for realizing equipment status monitoring and fault diagnosis.

[0116] By implementing the present invention, the following beneficial effects are achieved:

[0117] The present invention discloses a method and system for identifying the evolution law of multi-monitoring data based on waveform alignment. By performing curve fitting on the discrete data points after abnormal data processing and normalization processing of the monitoring data, a fitting curve is obtained. Perform correlation analysis on two measurement points within the fitting curve. When the correlation coefficient is higher than the preset threshold, it is considered that the correlation degree is high, and then calculate the displacement between the waveforms of the measurement points, so as to measure the correlation and delay between the two measurement points. By implementing the present invention, the correlation degree of the measurement points is preliminarily screened based on the correlation threshold, and then the curve alignment calculation is performed, which greatly improves the calculation efficiency of waveform alignment of the measurement point data, and encapsulates the identification method into a general method, solving the problem that it cannot be directly applied due to complex business logic and involving multi-layer data transformation processing.

[0118] It can be understood that the above embodiments only express the preferred implementation modes of the present invention, and the description is relatively specific and detailed, but it cannot be understood as a limitation to the scope of the patent of the present invention; it should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, the above technical features can be freely combined, and several deformations and improvements can also be made, which all belong to the protection scope of the present invention; therefore, all equivalent transformations and modifications made to the scope of the claims of the present invention should fall within the scope covered by the claims of the present invention.

Claims

1. A method for identifying the evolution law of multi-monitoring data based on waveform alignment, characterized in that It includes the following steps: S10: Obtain the monitoring data of nuclear power plant equipment and perform abnormal data processing on the monitoring data; S20: Perform normalization processing on the data after abnormal data processing; S30: Perform data curve fitting on the data after normalization processing to obtain a fitting curve; S40: Collect the fitting curve, perform correlation analysis on the pairwise measurement point combinations in the fitting curve. If the obtained correlation coefficient is higher than the preset threshold, calculate the displacement between the waveforms of the pairwise measurement points, and calculate the delay between the pairwise measurement points through the displacement; Step S20 includes: The normal operation state data includes the original data of multiple measurement points. Use the following formula to perform normalization processing on the normal operation state data: wherein is the measuring point serial number, is the data after normalization processing of the -th measuring point, is the original data of the -th measuring point, max( ) is the maximum value of , min( ) is the minimum value of ; Step S40 includes: Sample the multiple fitting curves formed by multiple measurement points. The sampling frequency is set to the maximum frequency in the original measurement point data to obtain the time-series data with consistent time points and frequencies for multiple measurement points; Use the Pearson similarity algorithm to calculate the pairwise combinations of the measurement points in the fitting curve to obtain the correlation coefficient; If the correlation coefficient of the measurement point combination [a, b] is higher than the preset threshold, then with measurement point a as the reference, within the time interval range of [-100, 100], translate the fitting curve of measurement point b 201 times respectively, and calculate the correlation coefficient Pn after each translation; if the calculated correlation coefficient Pn in a certain calculation is greater than the initial correlation coefficient P0, then retain this value; through the above calculation process, find the translation time interval n that makes the correlation coefficient reach the maximum value; the translation time interval n is the delay time between measurement point a and measurement point b; Based on the delay calculation results between all the pairwise measurement points, group the measurement points based on the correlation coefficient, divide the measurement points into relevant combinations, calculate the reference measurement points in the combination, and recalculate the delay of the measurement points according to the reference measurement points.

2. The method for identifying the evolution law of multi-monitoring data based on waveform alignment according to claim 1, wherein Step S10 further includes: S101: Obtain the time-series data sequence of each measurement point in the monitoring data, set the window range, and perform sliding calculation on all the original measurement point data according to the window range to calculate the average value and standard deviation of the data within the window range; S102: If the measurement point data at the center of the window range is not within the range set based on the average value and the standard deviation, then determine that the measurement point data is abnormal data; S103: Process the abnormal data using the median filtering algorithm.

3. The method for identifying the evolution law of multi-monitoring data based on waveform alignment according to claim 1, wherein Step S30 includes: Set a polynomial function, and use the least squares method to obtain the optimal parameter solution of the polynomial function, thereby obtaining the fitting curve.

4. The method for identifying the evolution law of multi-monitoring data based on waveform alignment according to claim 1, wherein The "if the obtained correlation coefficient is higher than the preset threshold" includes: If the absolute value of the obtained correlation coefficient is higher than the preset threshold; Among them, if the correlation coefficient is positive, it is determined that the change of the pairwise measurement points is positively correlated; if the correlation coefficient is negative, it is determined that the change of the pairwise measurement points is negatively correlated.

5. A multi-monitoring data evolution law recognition system based on waveform alignment, characterized in that It includes An abnormal data processing module, which is used to obtain the monitoring data of nuclear power plant equipment and perform abnormal data processing on the monitoring data; A normalization processing module for normalizing the data after abnormal data processing; A curve fitting module for performing data curve fitting on the data after normalization processing to obtain a fitting curve; A waveform alignment module for collecting the fitting curve, performing correlation analysis on pairwise measurement point combinations in the fitting curve. If the obtained correlation coefficient is higher than a preset threshold, calculate the displacement between the waveforms of the pairwise measurement points, and calculate the delay between the pairwise measurement points through the displacement; The normal operating state data includes the original data of multiple measurement points. The following formula is used to normalize the normal operating state data: where is the measuring point serial number, is the data after normalization processing of the th measuring point, is the original data of the th measuring point, max( ) is the maximum value of , min( ) is the minimum value of ; The waveform alignment module is used for: Sampling the multiple fitting curves formed by multiple measurement points, setting the sampling frequency as the maximum frequency in the original measurement point data, and obtaining the time-series data with consistent time points and frequencies for multiple measurement points; Using the Pearson similarity algorithm to calculate pairwise combinations of the measurement points in the fitting curve to obtain the correlation coefficient; If the correlation coefficient of the measurement point combination [a, b] is higher than the preset threshold, taking measurement point a as the reference, within the time interval range of [-100, 100], translate the fitting curve of measurement point b 201 times respectively, and calculate the correlation coefficient Pn after each translation. If the calculated correlation coefficient Pn in a certain calculation is greater than the initial correlation coefficient P0, then retain this value; through the above calculation process, find the translation time interval n that makes the correlation coefficient reach the maximum value; the translation time interval n is the delay time between measurement point a and measurement point b; Based on the calculation results of the delays between all the pairwise measurement points, group the measurement points based on the correlation coefficient, divide the measurement points into relevant combinations, calculate the reference measurement points in the combination, and recalculate the delays of the measurement points according to the reference measurement points.

6. The multi-monitoring data evolution law recognition system based on waveform alignment according to claim 5, characterized in that The abnormal data processing module further includes: A calculation unit for obtaining the time-series data sequence of each measurement point in the monitoring data, setting a window range, and performing sliding calculation on all the original measurement point data according to the window range to calculate the average value and standard deviation of the data within the window range; A judgment unit for determining that the measurement point data is abnormal data if the measurement point data at the center of the window range is not within the range set based on the average value and the standard deviation; A processing unit for processing the abnormal data using a median filtering algorithm.

7. The system for identifying the evolution law of multi-monitoring data based on waveform alignment according to claim 5, characterized in that, The curve fitting module includes: Setting a polynomial function, and obtaining the optimal parameter solution of the polynomial function by the least squares method, thereby obtaining the fitting curve.

8. The system for identifying the evolution law of multi-monitoring data based on waveform alignment according to claim 5, characterized in that, The case where if the obtained correlation coefficient is higher than the preset threshold includes: Analyzing whether the absolute value of the obtained correlation coefficient is higher than the preset threshold; Among them, if the correlation coefficient is positive, it is determined that the change of the pairwise measurement points is positively correlated; if the correlation coefficient is negative, it is determined that the change of the pairwise measurement points is negatively correlated.

Images