An abnormal value recognition method for relay parameters

Abstract

The present invention relates to the field of data processing, and particularly to a method for identifying abnormal values of relay parameters. The method includes: collecting historical normal two-dimensional parameters of the relay, obtaining a plurality of parameter clusters after clustering the historical normal two-dimensional parameters, and obtaining a fluctuation cluster after differentiating the two-dimensional parameters in the parameter clusters; collecting real-time two-dimensional parameters of the relay, where the real-time two-dimensional parameters are real-time current and real-time voltage, determining the corresponding parameter cluster and fluctuation cluster according to the time tag of the real-time two-dimensional parameters, and calculating a first abnormal degree and a second abnormal degree; calculating the overall abnormal degree Q of the relay, and when Q is greater than the threshold, there is an abnormality in the relay parameters. The present invention solves the problem that the prior art cannot significantly improve the accuracy of identifying abnormal values of relay parameters.

Description

Technical Field

[0001] The present invention relates to the field of data processing. More specifically, the present invention relates to a method for identifying outliers of relay parameters. Background Art

[0002] The research background of the method for identifying outliers of relay parameters based on big data mainly stems from the high requirements of the power system for safety and reliability. With the development of smart grids and the Internet of Things, a large amount of operation data has been generated in the application of relays in protection and control. By analyzing a large amount of relay data, the working state of the relay can be monitored in real time, outliers can be identified in a timely manner, so as to give early warnings of potential faults, reduce downtime and economic losses. At the same time, this method can also optimize maintenance strategies, improve the usage efficiency and lifespan of relays, and promote the development of intelligent operation and maintenance. Through in-depth research, the safety and stability of the power system can be further improved.

[0003] Currently, the patent application document with the publication number CN117454292A discloses a method and device for identifying outliers of relay parameters. This patent application document calculates the correlation matrix based on the dissimilarity matrix composed of the distances between parameters to determine the correlation between each parameter. And by normalizing the correlation matrix, the correlation probability matrix is obtained, so as to calculate the outlier probability of each parameter according to the correlation between parameters, making full use of the correlation between parameters. The situation of false detection is avoided, and the accuracy of outlier identification is improved. The solution in the above patent application document only considers the correlation between relay parameters and avoids false detection, but it is easy to cause missed detection, and its detection method is relatively simple and cannot significantly improve the accuracy of outlier identification of relay parameters. Summary of the Invention

[0004] To solve the problem that the accuracy of outlier identification of relay parameters cannot be significantly improved in the prior art, the present invention provides solutions in the following aspects.

[0005] The present invention provides a method for identifying outliers of relay parameters, including:

[0006] Collect historical normal two-dimensional parameters of the relay, cluster the historical normal two-dimensional parameters to obtain multiple parameter clusters, and perform differencing on the two-dimensional parameters in the parameter clusters to obtain a fluctuation cluster. The historical normal two-dimensional parameters are historical normal current and historical normal voltage, and the fluctuation cluster includes current fluctuations and voltage fluctuations; collect real-time two-dimensional parameters of the relay, the real-time two-dimensional parameters are real-time current and real-time voltage, determine the corresponding parameter cluster and fluctuation cluster according to the time tag of the real-time two-dimensional parameters, and calculate the first degree of abnormality and the second degree of abnormality , , where Represents the real-time current in the real-time two-dimensional parameters, represents the average current in the parameter cluster corresponding to the real-time two-dimensional parameters, represents the current variance in the parameter cluster corresponding to the real-time two-dimensional parameters, represents the real-time voltage in the real-time two-dimensional parameters, represents the average voltage in the parameter cluster corresponding to the real-time two-dimensional parameters, represents the voltage variance in the parameter cluster corresponding to the real-time two-dimensional parameters, represents the difference between the real-time current in the real-time two-dimensional parameters and the current collected at the previous moment, represents the average current fluctuation in the fluctuation cluster corresponding to the real-time two-dimensional parameters, represents the current fluctuation variance in the fluctuation cluster corresponding to the real-time two-dimensional parameters, represents the difference between the real-time voltage in the real-time two-dimensional parameters and the voltage collected at the previous moment, represents the average voltage fluctuation in the fluctuation cluster corresponding to the real-time two-dimensional parameters, represents the voltage fluctuation variance in the fluctuation cluster corresponding to the real-time two-dimensional parameters; calculates the overall abnormal degree of the relay , when is greater than the threshold, there is an abnormality in the relay parameters, where, , represents the degree of parameter chaos, represents the degree of fluctuation chaos, represents the first abnormal degree, represents the second abnormal degree; the degree of parameter chaos and the degree of fluctuation chaos respectively characterize the distribution of the parameter cluster and the fluctuation cluster corresponding to the real-time two-dimensional parameters.

[0007] This method performs clustering analysis on the historical normal two-dimensional parameters of the relay, identifies the normal parameter clusters and fluctuation clusters, and calculates two abnormal degree indicators based on the changes in the real-time two-dimensional parameters (current and voltage), so as to be able to monitor in real time whether the relay is abnormal. By performing clustering analysis on the two-dimensional parameters of historical normal current and voltage, multiple parameter clusters are formed. These clusters represent the normal current and voltage patterns under different working states. By taking the difference of the two-dimensional parameters in the parameter clusters, the fluctuation clusters are obtained, which can reflect the volatility of the current and voltage at different time points. The current and voltage of the relay are collected in real time, and by comparing with the historical parameter clusters and fluctuation clusters, it can be determined whether the real-time parameters fall within the normal range and whether the real-time fluctuation conditions are normal. When calculating the first abnormal degree and the second abnormal degree When comparing the real-time current and voltage with the mean and variance in the parameter cluster, the fluctuations of the relay outside the normal range can be accurately identified. If the real-time data deviates far from the mean of the cluster, it indicates that there may be a risk of relay abnormality or failure. By comparing the differences between the real-time current and voltage and the data at the previous moment, and combining the mean and variance in the fluctuation cluster, sudden changes or abnormal fluctuations of the relay can be effectively detected. This can timely detect sudden changes in the relay state, especially those short-term fluctuations that may lead to failures. Calculating and monitoring the first and second degrees of abnormality in real time can help the operation and maintenance personnel quickly locate the reasons for the relay abnormality. For example, if the degree of current abnormality is high while the voltage change is small, it may be a failure caused by uneven current load; if the voltage fluctuates abnormally, it may be a power supply problem. Combining the automatic calculation of the first and second degrees of abnormality can realize the intelligent monitoring of the relay, making the monitoring and analysis of the relay operation state more efficient and accurate. The advantage of this method is that through the comprehensive monitoring of the relay working state, combining clustering analysis and fluctuation analysis, it can not only accurately judge whether the relay is abnormal, but also give early warnings, thereby reducing the probability of failures and ensuring the reliability and stability of the relay.

[0008] Preferably, after collecting the historical normal two-dimensional parameters of the relay, it includes: performing missing data filling processing on the historical normal two-dimensional parameters.

[0009] By filling in the missing data, the integrity of the historical normal two-dimensional parameter data is ensured, avoiding incorrect clustering analysis or subsequent calculation deviations caused by incomplete data.

[0010] Preferably, after clustering the historical normal two-dimensional parameters to obtain multiple parameter clusters, it includes: using the method of clustering ordered samples for clustering.

[0011] The method of clustering ordered samples can consider the sequentiality of time in the data, which is very important for the historical normal two-dimensional parameters of the relay (such as current and voltage). The working state of the relay usually has time continuity. Ignoring this may lead to inaccurate clustering results, thereby affecting the subsequent abnormal detection effect. By taking the time series data as a key factor, the method of clustering ordered samples effectively improves the timeliness of the clustering results, making the parameter clusters after clustering better reflect the actual operation state of the relay.

[0012] Preferably, the parameter chaos degree and the fluctuation chaos degree are specifically: the sum of squared errors within the parameter cluster and the sum of squared errors within the fluctuation cluster.

[0013] Preferably, after calculating the overall abnormality degree of the relay, it includes: when the overall abnormality degree is less than the threshold, dividing the real-time two-dimensional parameters into the historical normal two-dimensional parameters and dynamically updating the historical normal two-dimensional parameters.

[0014] The dynamic update of historical parameters can adjust the normal working range in real time according to the latest operating status of the relay. When the relay has been in a normal operating state for a period of time and the degree of abnormality is less than the threshold, the real-time data can become the new "normal" data. This dynamic update enables the system to better adapt to the possible long-term changes of the relay and improves the adaptability and flexibility of the model.

[0015] Preferably, when After the relay parameters are abnormal when greater than the threshold, it further includes: performing abnormal warning through the color of the fault indicator light and / or the sound of the fault warning bell.

[0016] The beneficial effects of the present invention are as follows: By performing clustering analysis on the two-dimensional parameters of historical normal current and voltage, multiple parameter clusters are formed. These clusters represent the normal current and voltage patterns under different operating states. By taking the difference of the two-dimensional parameters in the parameter clusters, a fluctuation cluster is obtained, which can reflect the volatility of the current and voltage at different time points. By collecting the current and voltage of the relay in real time and comparing them with the historical parameter clusters and the fluctuation cluster, it can be determined whether the real-time parameters fall within the normal range and whether the real-time fluctuation condition is normal. The advantage of this method is that through the comprehensive monitoring of the relay operating state, combining clustering analysis and fluctuation analysis, it can accurately judge whether the relay is abnormal and can also give early warnings. Description of the Drawings

[0017] By reading the following detailed description with reference to the drawings, the above and other objects, features, and advantages of the exemplary embodiments of the present invention will become readily understandable. In the drawings, several embodiments of the present invention are shown in an exemplary rather than restrictive manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0018] Figure 1 is a flowchart of a method for identifying outliers of relay parameters provided by an embodiment of the present invention. Detailed Embodiments

[0019] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0020] The following will describe the specific embodiments of the present invention in detail with reference to the drawings.

[0021] Figure 1It is a flowchart of a method for identifying abnormal values of relay parameters according to an embodiment of the present invention, including the following steps:

[0022] S101. Collect historical normal two-dimensional parameters of the relay, obtain multiple parameter clusters after clustering the historical normal two-dimensional parameters, and obtain a fluctuation cluster after differentiating the two-dimensional parameters in the parameter clusters.

[0023] The historical normal two-dimensional parameters are historical normal current and historical normal voltage, and the fluctuation cluster includes current fluctuation and voltage fluctuation.

[0024] In one embodiment, during the operation of the relay, the electricity consumption demands at different stages within a day are different. Therefore, the parameter abnormality situations at different stages are also different. For example, the parameters that are normal during the peak electricity consumption period may be abnormal during the low electricity consumption period. If the same standard is directly used to define abnormalities for the parameter values at all times, the detection accuracy of abnormalities is relatively low. Therefore, ordered sample clustering can be adopted for the historical normal two-dimensional parameters of the relay. The ordered sample clustering method effectively improves the timeliness of the clustering result by taking the time series data as a key factor.

[0025] The collection of the historical normal two-dimensional parameters of the relay includes: collecting the historical normal two-dimensional parameters and real-time two-dimensional parameters through a current sensor and a voltage sensor.

[0026] For example, the current sensor and the voltage sensor collect relay parameters every ten seconds. The current and voltage collected in the past ten days are all normal parameters. All the current and voltage data for these ten days are used as the historical normal two-dimensional parameters. Then, according to the ordered sample clustering algorithm, the historical normal two-dimensional parameters are clustered. After clustering, these data can be divided into multiple clustering clusters according to the time tags. For example, 0:00 to 8:00, 11:00 to 1:00, 17:00 to 19:00, and 22:00 to 24:00 form a parameter cluster, called the peak cluster, and the other time is another parameter cluster, the low peak cluster. Of course, the ordered sample clustering algorithm can also be clustered into multiple other clusters, which will not be elaborated here.

[0027] After obtaining the parameter cluster, the fluctuation cluster can be obtained by taking the difference according to the time sequence of the parameters in the parameter cluster. For example, in a parameter cluster, there are five two-dimensional parameters: (3A, 5V), (3.2A, 5.1V), (3.0A, 5.1V), (3.3A, 5.2V), and (2.9A, 5.1V), and the corresponding time tags are 15:10:00, 15:10:10, 15:10:20, 15:10:30, and 15:10:40. Then, after taking the difference according to the time sequence, four fluctuation arrays can be obtained: (0.2A, 0.1V), (-0.2A, 0V), (0.3A, 0.1V), and (-0.4A, -0.1V). The first data in the fluctuation array is the current fluctuation, and the second data is the voltage fluctuation. The four fluctuation arrays together constitute the fluctuation cluster corresponding to the parameter cluster.

[0028] Since the sensor may be affected by the external environment and occasionally fail to collect the current and voltage during the process of collecting the current and voltage, missing values will occur. Therefore, after collecting the historical normal two-dimensional parameters of the relay, it is necessary to perform missing value filling processing on the historical normal two-dimensional parameters, and generally interpolation filling can be used for processing.

[0029] S102. Collect the real-time two-dimensional parameters of the relay. The real-time two-dimensional parameters are the real-time current and the real-time voltage. Determine the corresponding parameter cluster and fluctuation cluster according to the time tag of the real-time two-dimensional parameters, and calculate the first abnormal degree and the second abnormal degree.

[0030] In some embodiments, calculate the first abnormal degree and the second abnormal degree , specifically: , where represents the real-time current in the real-time two-dimensional parameters, represents the average current in the parameter cluster corresponding to the real-time two-dimensional parameters, represents the variance of the current in the parameter cluster corresponding to the real-time two-dimensional parameters, represents the real-time voltage in the real-time two-dimensional parameters, represents the average voltage in the parameter cluster corresponding to the real-time two-dimensional parameters, represents the variance of the voltage in the parameter cluster corresponding to the real-time two-dimensional parameters, represents the difference between the real-time current in the real-time two-dimensional parameters and the current collected at the previous moment, represents the average current fluctuation in the fluctuation cluster corresponding to the real-time two-dimensional parameters, represents the variance of the current fluctuation in the fluctuation cluster corresponding to the real-time two-dimensional parameters, represents the difference between the real-time voltage in the real-time two-dimensional parameters and the voltage collected at the previous moment, Represents the average voltage fluctuation in the corresponding fluctuation cluster in the real-time two-dimensional parameters. Represents the variance of voltage fluctuation in the corresponding fluctuation cluster in the real-time two-dimensional parameters.

[0031] For example, in the parameter cluster obtained in S101, the period from 0 o'clock to 8 o'clock is the peak cluster. The time tag of the real-time two-dimensional parameter is 06:25:20, which is within the period from 0 o'clock to 8 o'clock. Therefore, it corresponds to the peak cluster from 0 o'clock to 8 o'clock. When calculating the first abnormal degree of this real-time two-dimensional parameter, the average current, current variance, average voltage, and voltage variance in this peak cluster are used for calculation. The two-dimensional parameters in this peak cluster are differenced according to the time sequence to obtain the corresponding peak fluctuation cluster. When calculating the second abnormal degree of this real-time two-dimensional parameter, the average current fluctuation, current fluctuation variance, average voltage fluctuation, and voltage fluctuation variance in this peak fluctuation cluster are used for calculation.

[0032] S103. Calculate the overall abnormal degree Q of the relay. When Q is greater than the threshold, there is an abnormality in the relay parameters.

[0033] When actually calculating the overall abnormal degree Q, generally, the sum of the chaos degree of the relay parameters and the chaos degree of the parameter fluctuation is constant. The first abnormal degree usually occupies a greater weight. Since the difference calculation will amplify the volatility of the data, the chaos degree of the fluctuation in the fluctuation cluster is usually larger than the chaos degree of the parameters in the parameter cluster So in some embodiments, the overall abnormal degree , represents the chaos degree of the parameters, represents the chaos degree of the fluctuation, represents the first abnormal degree, represents the second abnormal degree; the chaos degree of the parameters and the chaos degree of the fluctuation respectively characterize the distribution of the parameter cluster and the fluctuation cluster corresponding to the real-time two-dimensional parameters.

[0034] The chaos degree of the parameters and the chaos degree of the fluctuation are specifically: the sum of the squared errors within the parameter cluster and the sum of the squared errors within the fluctuation cluster. It should be noted that calculating the sum of the squared errors within the clustering cluster is a well-known technology and will not be elaborated here.

[0035] In addition, in order to dynamically update the historical normal two-dimensional parameters and improve adaptability and flexibility, after calculating the overall abnormal degree of the relay, it further includes: when the overall abnormal degree is less than the threshold, dividing the real-time two-dimensional parameters into the historical normal two-dimensional parameters and dynamically updating the historical normal two-dimensional parameters.

[0036] When the relay parameters are abnormal, in order to quickly convey the abnormal state of the relay, in some embodiments, when the overall abnormal degree When there is an abnormality in the relay parameters when they are greater than the threshold, it further includes: giving an abnormality warning through the color of the fault indicator light and / or the sound of the fault warning bell.

[0037] The present invention performs clustering analysis on the two-dimensional parameters of historical normal current and voltage to form multiple parameter clusters. These clusters represent the normal current and voltage patterns under different working states. By taking the difference of the two-dimensional parameters in the parameter clusters, a fluctuation cluster is obtained, which can reflect the fluctuations of the current and voltage at different time points. The current and voltage of the relay are collected in real time, and by comparing with the historical parameter clusters and the fluctuation cluster, it can be determined whether the real-time parameters fall within the normal range and whether the real-time fluctuation condition is normal. The advantage of this method is that through the comprehensive monitoring of the working state of the relay, combining clustering analysis and fluctuation analysis, it can accurately judge whether the relay is abnormal, and can also give an early warning, significantly improving the problem of the accuracy rate of identifying abnormal values of relay parameters.

[0038] In the description of this specification, the meanings of "a plurality of" and "several" are at least two, such as two, three or more, etc., unless otherwise specifically and clearly defined.

[0039] Although this specification has shown and described multiple embodiments of the present invention, it is obvious to those skilled in the art that such embodiments are provided only by way of example. Those skilled in the art will think of many changes, alterations and alternative ways without departing from the spirit and idea of the present invention. It should be understood that various alternative solutions to the embodiments of the present invention described herein can be adopted in the practice of the present invention.

Claims

1. A method for identifying abnormal values ​​of relay parameters, characterized in that: include: Collecting historical normal two-dimensional parameters of the relay, clustering the historical normal two-dimensional parameters using an ordered sample clustering method to obtain multiple parameter clusters, and differentiating the two-dimensional parameters in the parameter clusters to obtain a fluctuation cluster, wherein the historical normal two-dimensional parameters are historical normal current and historical normal voltage, and the fluctuation cluster includes current fluctuation and voltage fluctuation; Collect the real-time two-dimensional parameters of the relay, which are the real-time current and real-time voltage, determine the corresponding parameter cluster and fluctuation cluster according to the time label of the real-time two-dimensional parameters, and calculate the first abnormality degree and the second degree of abnormality , ,in, Represents the real-time current in real-time two-dimensional parameters, Represents the current mean value in the parameter cluster corresponding to the real-time two-dimensional parameter, represents the current variance in the parameter cluster corresponding to the real-time two-dimensional parameter, Represents the real-time voltage in real-time two-dimensional parameters, Represents the voltage mean value in the parameter cluster corresponding to the real-time two-dimensional parameter, Represents the voltage variance in the parameter cluster corresponding to the real-time two-dimensional parameter, It represents the difference between the real-time current in the real-time two-dimensional parameters and the current collected at the previous moment. represents the mean current fluctuation in the corresponding fluctuation cluster in the real-time two-dimensional parameter, represents the current fluctuation variance in the corresponding fluctuation cluster in the real-time two-dimensional parameter, It represents the difference between the real-time voltage in the real-time two-dimensional parameters and the voltage collected at the previous moment. Represents the mean voltage fluctuation in the fluctuation cluster corresponding to the real-time two-dimensional parameter, Represents the voltage fluctuation variance in the corresponding fluctuation cluster in the real-time two-dimensional parameter; Calculate the overall abnormality of the relay ,when When the value is greater than the threshold, the relay parameters are abnormal, among which, , Indicates the degree of parameter confusion, Indicates the degree of fluctuation disorder. Indicates the first abnormality level, Indicates the second abnormality degree; the parameter chaos degree and the fluctuation chaos degree respectively characterize the distribution of the parameter cluster and the fluctuation cluster corresponding to the real-time two-dimensional parameters.

2. The method for identifying abnormal values ​​of relay parameters according to claim 1, characterized in that: After collecting the historical normal two-dimensional parameters of the relay, the method includes: performing missing completion processing on the historical normal two-dimensional parameters.

3. The method for identifying abnormal values ​​of relay parameters according to claim 1, characterized in that: The parameter disorder degree and the fluctuation disorder degree are specifically: the intra-cluster error sum of the parameter cluster and the intra-cluster error sum of the fluctuation cluster.

4. The method for identifying abnormal values ​​of relay parameters according to claim 1, characterized in that: The collecting of the historical normal two-dimensional parameters of the relay includes: collecting the historical normal two-dimensional parameters and the real-time two-dimensional parameters through a current sensor and a voltage sensor.

5. The method for identifying abnormal values ​​of relay parameters according to claim 1, characterized in that: After calculating the overall abnormality degree of the relay, the method includes: when the overall abnormality degree is less than a threshold value, dividing the real-time two-dimensional parameters into historical normal two-dimensional parameters, and dynamically updating the historical normal two-dimensional parameters.

6. The method for identifying abnormal values ​​of relay parameters according to claim 1, characterized in that: When When the relay parameter is abnormal when it is greater than the threshold, it also includes: abnormal warning through the color of the fault prompt light and / or the sound of the fault prompt bell.

Images