A monitoring data processing method for electric power instruments

Abstract

The application relates to the technical field of data cleaning, and particularly relates to a power instrument monitoring data processing method, which comprises the following steps: acquiring monitoring data of a power instrument, and determining parameter change curves of different dimensions according to the monitoring data; determining power consumption trend characteristics of each time sequence point by using the parameter change curves of different dimensions, and then determining parameter change characteristic values of each time sequence point; determining parameter detection disturbance degrees under different dimensions by using the parameter change characteristic values; acquiring dimension distances of at least two to-be-measured data points under different dimensions by using the parameter detection disturbance degrees, determining a data monitoring result according to the dimension distances, and then cleaning abnormal data according to the monitoring result; that is, the application reduces the interference of power data change on abnormal detection by analyzing power consumption data change characteristics, improves detection accuracy, and further improves the accuracy of power system analysis.

Description

Technical Field

[0001] This invention relates to the technical field of data cleaning, and more specifically to a method for processing monitoring data from electrical instruments. Background Technology

[0002] Electrical instruments are specialized devices used to measure, monitor, and analyze various parameters of power systems. They play a vital role in power engineering, ensuring the stable operation of power systems, assessing power quality, and monitoring equipment performance. Modern power systems are relatively complex, containing numerous control components. To ensure the system's rational and reliable operation, electrical instruments are needed to monitor each component and, combined with monitoring data, to evaluate the power system's operational status.

[0003] In the actual monitoring of power systems by power instruments, power parameter information from various aspects of the power system is collected and transformed into data for evaluation and analysis. However, with prolonged use, the data acquisition process is susceptible to electrical noise, leading to abnormal monitoring data. Power system parameters change as the system operates, and existing anomaly detection algorithms often overlook this change, resulting in detection results that fail to meet the data cleaning requirements for power system monitoring. Summary of the Invention

[0004] To address the technical problem of abnormal monitoring data, the present invention aims to provide a method for processing monitoring data of power instruments, which can effectively clean abnormal monitoring data.

[0005] The specific technical solution adopted is as follows: a method for processing monitoring data of power instruments is provided, the method comprising: acquiring monitoring data of power instruments, and determining parameter change curves in different dimensions based on the monitoring data; using the parameter change curves in different dimensions to determine the electricity consumption trend characteristics at each time point, and then determining the parameter change characteristic value at each time point; using the parameter change characteristic value to determine the parameter detection disturbance degree in different dimensions; using the parameter detection disturbance degree to obtain the dimensional distance between at least two data points to be measured in different dimensions, determining the data monitoring result based on the dimensional distance, and then cleaning abnormal data based on the monitoring result.

[0006] In one embodiment of this application, the step of acquiring monitoring data from a power instrument and determining parameter change curves in different dimensions based on the monitoring data includes: using a power instrument to detect a target node to acquire the monitoring data, wherein the monitoring data includes monitoring data in different dimensions; and using the monitoring data in different dimensions to perform curve fitting to acquire parameter change curves in different dimensions.

[0007] In one embodiment of this application, determining the electricity consumption trend characteristics of each time series point using the parameter change curves of different dimensions, and then determining the parameter change characteristic value of each time series point, includes: determining an electricity consumption fitting curve using the parameter change curves of different dimensions; obtaining the derivative of each time series point on the electricity consumption fitting curve, and using the derivative to determine the mean derivative value of each time series point on the electricity consumption fitting curve; determining the electricity consumption trend characteristics of each time series point using a preset activation function, the derivative, and the mean derivative value; obtaining the relative time series position of each time series point in a preset interval of the parameter change curves of different dimensions, and then using the relative time series position and the electricity consumption trend characteristics to determine the parameter change characteristic value of each time series point.

[0008] In one embodiment of this application, the parameter change curve includes a current change curve and a voltage change curve; obtaining the relative time position of each time point in a preset interval of the parameter change curve in different dimensions, and then using the relative time position and the electricity consumption trend feature to determine the parameter change feature value of each time point, includes: obtaining the first relative time position of each time point in the preset interval of the current change curve, and obtaining the second relative time position of each time point in the preset interval of the voltage change curve; obtaining the first derivative of each time point in the current change curve, and obtaining the second derivative of each time point in the voltage change curve, wherein the first derivative and the second derivative are obtained after standardization; and using the first relative time position, the second relative time position, the first derivative, the second derivative, and the electricity consumption trend feature to determine the parameter change feature value of each time point.

[0009] In one embodiment of this application, determining the parameter change characteristic value at each time point using the first relative time position, the second relative time position, the first derivative, the second derivative, and the electricity consumption trend characteristics includes: determining the overall change characteristics of power parameters using the first relative time position, the second relative time position, and the electricity consumption trend characteristics; determining data change characteristics using the first relative time position, the second relative time position, the first derivative, and the second derivative; and determining the parameter change characteristic value at each time point using the overall change characteristics of power parameters and the data change characteristics.

[0010] In one embodiment of this application, the parameter change curve includes a current change curve and a voltage change curve; determining the parameter detection perturbation degree in different dimensions using the parameter change feature values ​​includes: obtaining a first derivative time series of the current change curve and a second derivative time series of the voltage change curve; obtaining the monitoring data volume in each dimension, obtaining the average difference between the monitoring data of the target time point and adjacent time points in each dimension, and obtaining the average difference between the parameter change feature values ​​of the target time point and adjacent time points; and determining the parameter detection perturbation degree in each dimension using the first derivative time series, the second derivative time series, the monitoring data volume, the average difference, and the average difference.

[0011] In one embodiment of this application, determining the parameter detection perturbation degree of each dimension using the first derivative time series, the second derivative time series, the amount of monitoring data, the mean difference, and the mean difference includes: determining the data change frequency parameter of the target time series point using the mean difference and the mean difference, and then determining the first calculation parameter of each dimension using the data change frequency parameter and the amount of monitoring data; determining the derivative time series coefficients of the voltage change curve and the current change curve using the first derivative time series and the second derivative time series; and determining the parameter detection perturbation degree of each dimension using a preset exponential function, the first calculation parameter, and the derivative time series coefficients.

[0012] In one embodiment of this application, the step of using the parameter to detect perturbation degree, obtaining the dimensional distance between at least two data points to be tested in different dimensions, determining the data monitoring result based on the dimensional distance, and then cleaning abnormal data based on the monitoring result includes: obtaining data values ​​of at least two data points to be tested in different dimensions, and obtaining parameter change feature values ​​of at least two data points to be tested; using the data values, the parameter change feature values, and the parameter detection perturbation degree, determining the dimensional distance between at least two data points to be tested in different dimensions; using the dimensional distance in different dimensions, identifying abnormal data, and then determining the data monitoring result.

[0013] In one embodiment of this application, a first average data value of the data points contained in each monotonic interval of the current change curve is obtained, and a second average data value of the data points contained in each monotonic interval of the voltage change curve is obtained; the data monitoring results are optimized using the first average data value and the second average data value.

[0014] In one embodiment of this application, the step of cleaning abnormal data based on monitoring results includes: deleting the abnormal data from the monitoring results; and then filling it in chronological order using interpolation.

[0015] This invention offers the following advantages: By establishing a fitting curve for monitoring data and analyzing the overall trend of data changes, it provides parameters for the anomaly detection and judgment process. Then, by combining the electricity consumption change trend and the characteristics of current and voltage curves, it analyzes the detection disturbance degree of the two-dimensional data, further improving the accuracy of data change characteristic analysis while evaluating the interference characteristics of detection, providing parameters for subsequent adjustments to the anomaly detection process. Finally, it optimizes the distance measurement method by combining the disturbance degree and data change characteristics of each dimension of data, and optimizes the anomaly detection algorithm based on the change characteristics of the fitting curve. Compared with existing technologies, the detection process adds analysis of electricity consumption data change characteristics, reduces the interference of power data changes on anomaly detection, improves detection accuracy, and further increases the accuracy of power system analysis. Attached Figure Description

[0016] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating a monitoring data processing method for an electrical instrument, as provided in one embodiment of the present invention. Detailed Implementation

[0018] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a monitoring data processing method for an electrical instrument according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0020] The specific scheme of the monitoring data processing method for an electrical instrument provided by the present invention will be described in detail below with reference to the accompanying drawings.

[0021] Please see Figure 1 The diagram illustrates a flowchart of a monitoring data processing method for an electrical instrument according to an embodiment of the present invention.

[0022] like Figure 1 As shown, a method for processing monitoring data from an electrical instrument includes the following steps:

[0023] S10. Obtain monitoring data from the power instrument and determine parameter change curves in different dimensions based on the monitoring data.

[0024] Among them, power instruments are specialized equipment used to measure, monitor and analyze various parameters of the power system, and are used to ensure the stable operation of the power system, assess power quality, and monitor equipment performance; monitoring data refers to the data obtained by measuring various power consumption parameters of the power system through power instruments; different dimensions may include current dimension, voltage dimension, power consumption dimension, etc.

[0025] Specifically, different nodes of the power system are tested using electrical instruments to obtain various power consumption parameters corresponding to different nodes, such as current, voltage, and power consumption, as data in different dimensions; then, current change curves are determined based on the changes in current over time, voltage change curves are determined based on the changes in voltage over time, and power consumption change curves are determined based on the changes in power consumption over time.

[0026] S20. Using parameter change curves of different dimensions, determine the electricity consumption trend characteristics at each time point, and then determine the parameter change characteristic value at each time point.

[0027] The parameter change curve for each dimension can include many time points based on the division of the time line. Therefore, the electricity consumption trend characteristics at each time point can be determined based on the parameter change curves corresponding to each dimension. The parameter change characteristic value refers to the comprehensive change characteristics of the power parameters of each dimension corresponding to the time point.

[0028] Specifically, after obtaining parameter change curves in different dimensions, the parameters corresponding to each time series point can be determined from the parameter change curves based on the division of time series points according to the timeline. This serves as the electricity consumption trend characteristic of each time series point. Then, the parameter changes in each dimension of each time series point are analyzed to determine the parameter change characteristic value of each time series point.

[0029] S30. Utilize parameter change characteristic values ​​to determine the parameter detection perturbation degree under different dimensions.

[0030] Among them, parameter detection disturbance degree refers to the data change reflected by power parameter fluctuations.

[0031] Specifically, after obtaining the parameter change feature value of each time series point, the parameter change feature value corresponding to the current time series point and the adjacent time series points is used to determine the frequency of change of each dimension of data, and then the parameter detection perturbation degree under different dimensions is obtained.

[0032] S40. Utilize parameters to detect disturbance degree, obtain the dimensional distance between at least two data points to be tested in different dimensions, determine the data monitoring result based on the dimensional distance, and then clean the abnormal data based on the monitoring result.

[0033] Among them, dimensional data refers to the distances corresponding to different time points in the data space, which is used to reflect the local density in the data space.

[0034] Specifically, after obtaining the parameter detection disturbance degree under different dimensions, the dimensional data between different test data points are obtained by using the data values ​​of different test data points under different dimensions and the corresponding power parameter change characteristic values ​​and corresponding parameter detection disturbance degrees. This can highlight the distance between abnormal data and normal data, filter out abnormal data, obtain data monitoring results, and then clean the abnormal data.

[0035] In this embodiment, by adding analysis of the characteristics of changes in electricity consumption data during the detection process, the interference of changes in electricity data on anomaly detection is reduced, the detection accuracy is improved, and the accuracy of power system analysis is further increased.

[0036] In some embodiments, step S10, which involves acquiring monitoring data from the power instrument and determining parameter change curves in different dimensions based on the monitoring data, may further include the following operations:

[0037] First, electrical instruments are used to detect the target node in order to obtain monitoring data, which includes monitoring data from different dimensions.

[0038] The target node refers to a node in the power system that can be monitored.

[0039] Specifically, power instruments are connected to the target node of the power system to obtain various power parameters corresponding to the target node, such as current, voltage, and power consumption; and to obtain historical monitoring data during the detection period, as well as various power consumption parameters corresponding to each time point.

[0040] Next, curve fitting was performed using monitoring data from different dimensions to obtain parameter change curves from different dimensions.

[0041] Because power data changes with the changes in electrical equipment within the power system during operation, it is necessary to fit the various power data acquired by the power instruments into a curve in order to analyze the trend characteristics of data changes. The horizontal axis of the curve can be the detection time, and the vertical axis can be the current dimension data value.

[0042] Specifically, the curve corresponding to the current can be a current change curve, with the horizontal axis representing the current detection time and the vertical axis representing the current detection value; the curve corresponding to the voltage can be a voltage change curve, with the horizontal axis representing the voltage detection time and the vertical axis representing the voltage detection value; and the curve corresponding to the electricity consumption can be an electricity consumption change curve.

[0043] It is understandable that the data detected by the current change curve and the voltage change curve are the current values. Therefore, the electricity consumption change curve can be determined by the current change curve and the voltage change curve. The data detected by the electricity consumption change curve is usually the cumulative value.

[0044] In some embodiments, step S20 uses parameter change curves of different dimensions to determine the electricity consumption trend characteristics at each time point, and then determines the parameter change characteristic value at each time point, which may include the following operations:

[0045] First, the power consumption change curve is determined using parameter change curves of different dimensions; among them, the parameter change curves include the current change curve and the voltage change curve.

[0046] As mentioned above, the data detected by the current change curve and the voltage change curve are the current values. Therefore, the electricity consumption change curve can be determined by the current change curve and the voltage change curve. The data detected by the electricity consumption change curve is usually the cumulative value.

[0047] Next, the derivative of each time point on the electricity consumption change curve is obtained, and the mean value of the derivative of each time point on the electricity consumption change curve is determined using the derivative.

[0048] Specifically, if the electricity consumption change curve has multiple time points, including the t-th time point, then the derivative p of the t-th time point on the electricity consumption change curve is obtained. t ′ This allows us to obtain the mean of the derivatives corresponding to multiple time series points.

[0049] Next, using the preset activation function, derivative, and mean derivative, the electricity consumption trend characteristics at each time point are determined;

[0050] The monitoring data in the power system changes with the changes in electrical equipment, and these changes are mainly reflected in the overall electricity consumption of the power system. Therefore, the current electricity consumption trend can be analyzed by combining the changes in the electricity consumption curve. Commonly used activation functions can be used, such as sigmoid(.).

[0051] Specifically, by calculating the difference between the derivative and the mean of the derivatives at each time point using a preset activation function, the electricity consumption trend characteristics corresponding to each time point can be determined. However, directly using the derivative is insufficient to reflect the electricity consumption trend characteristics between different time points. To distinguish the electricity consumption trend characteristics between different time points, it is necessary to calculate the difference between the electricity consumption characteristics at each time point and the overall electricity consumption characteristics. The mean of the derivatives of the electricity consumption curve represents the overall characteristics of electricity consumption, and then the difference between the electricity consumption at a certain point and the overall electricity consumption characteristics is calculated. This reflects the electricity consumption trend at this specific time point compared to overall electricity consumption data. Finally, the difference result is placed between (0, 1) using the sigmoid function, and β is used... t This represents the electricity consumption trend characteristic at time point t. A larger value for the electricity consumption trend characteristic indicates that the electricity consumption at the current time point is greater compared to the overall electricity consumption situation.

[0052] Electricity consumption trend characteristics are calculated as follows:

[0053]

[0054] Where, β t To represent electricity consumption trend characteristics, sigmoid(.) is the preset activation function, p t ′ Let be the derivative of the electricity consumption change curve at the t-th time point. This represents the average derivative of each time point on the electricity consumption change curve.

[0055] It is understandable that after differentiating the fitted curve of electricity consumption, its derivative p t ′ This can represent the power consumption at the t-th time point.

[0056] Then, the relative time position of each time point in the preset interval of the parameter change curve in different dimensions is obtained, and the parameter change characteristic value of each time point is determined by using the relative time position and electricity consumption trend characteristics.

[0057] After obtaining the electricity consumption trend characteristics of the power system, the actual monitored power parameters can be analyzed. During periods of rapid change in electricity consumption trends, electrical noise that affects the accuracy of power instrument detection is more likely to occur in the power system. Furthermore, to ensure the stability of the power system during operation, power parameters are typically not frequently adjusted. To increase the accuracy of the power data analysis process, after obtaining the electricity consumption trend characteristics, it is necessary to further analyze the detection disturbance degree by combining the changing characteristics of the power monitoring data values.

[0058] Specifically, the voltage and current data are standardized using the range standardization method; the monotonic intervals of the current and voltage change curves are obtained using the derivative values; and to analyze the monitoring disturbance of data in different dimensions, it is first necessary to analyze the characteristic values ​​of power parameter changes in each dimension of the monitoring data points.

[0059] Obtain the first relative timing position of each timing point in a preset interval of the current change curve, and obtain the second relative timing position of each timing point in a preset interval of the voltage change curve.

[0060] The preset interval in the current change curve can be the increasing or decreasing interval in the current change curve, that is, the monotonic interval; the preset interval in the voltage change curve can be the increasing or decreasing interval in the voltage change curve, that is, the monotonic interval.

[0061] Specifically, taking the t-th time point as an example, the first relative time position of the t-th time point within a preset interval in the current change curve is obtained. And obtain the second relative timing position of the t-th timing point within a preset interval in the voltage change curve.

[0062] Next, the first derivative of each time point in the current change curve and the second derivative of each time point in the voltage change curve are obtained, wherein the first derivative and the second derivative are obtained after standardization.

[0063] The first derivative can be the derivative value obtained by range normalization of the current change curve at each time point; the second derivative can be the derivative value obtained by range normalization of the voltage change curve at each time point.

[0064] Specifically, taking the t-th time series point as an example, the first derivative ||I|| is obtained by standardizing the range in the current change curve at the t-th time series point. t ′ ‖, and obtain the second derivative ‖U obtained by range normalization of the voltage change curve at the t-th time point. t ′ ‖.

[0065] Next, using the first relative time position, the second relative time position, the first derivative, the second derivative, and the electricity consumption trend characteristics, the parameter change characteristic value of each time point is determined.

[0066] Specifically, taking the t-th time series point as an example, the electricity consumption trend feature β corresponding to the t-th time series point is used. t First relative timing position Second relative timing position First derivative ||I t′ ||、Second derivative|U t ′ ||, to determine the parameter change characteristic value at the t-th time point.

[0067] Among them, the first relative timing position can be used Second relative timing position Electricity consumption trend characteristics β t Determine the overall variation characteristics of power parameters; utilize the first relative time sequence position Second relative timing position First derivative ||I t ′ || and second derivative|U t ′ The characteristics of data change are determined; then, based on the overall characteristics of power parameter change and the characteristics of data change, the characteristic values ​​of parameter change at each time point are determined.

[0068] The characteristic values ​​of parameter changes are calculated as follows:

[0069]

[0070] Where, d t β represents the characteristic value of parameter change at the t-th time point. t This represents the electricity consumption trend characteristics at time point t. This represents the first relative timing position of the t-th timing point within a preset interval in the current variation curve. This represents the second relative timing position of the t-th timing point within a preset interval in the voltage change curve; ||I t ′ || represents the first derivative obtained by range normalization at the t-th time point in the current variation curve, ||U t ′ ‖ represents the second derivative obtained by range normalization of the voltage change curve at the t-th time point.

[0071] It is understandable that the first relative timing position Second relative timing position The calculation method is as follows Among them, t M ,t m These are the maximum and minimum time series values ​​of the increment / decrement interval to which the t-th time series point belongs, respectively.

[0072] Understandably, to represent the changing characteristics of the t-th time series point, in addition to the electricity consumption trend characteristics reflected in the electricity consumption change curve, it is also necessary to analyze the changing characteristics of the monitoring data. The electricity consumption trend characteristics are the result data obtained from further analysis of current and voltage; the changes in this data depend on the changes in current and voltage, reflecting the overall characteristics of electricity consumption. However, for anomaly monitoring, analyzing the data change characteristics of a specific time series point solely through the result data is insufficient to resist the influence of abnormal data. The electricity change process also needs to be reflected by its range of change: data at the end of the monotonic interval of the data series has a weaker ability to reflect the changing characteristics of electricity parameters. In addition to the overall electricity consumption trend reflected by the result data (i.e., electricity consumption), the degree of change also needs to be reflected by two actual monitoring data points.

[0073] To further improve the accuracy of the variation characteristics at the t-th time point and enhance its noise resistance, within the monotonic interval of the fitted curve corresponding to a certain monitoring data, The larger the value, the more pronounced the trend of the power parameter change characteristic value in the time series direction at the t-th time point is about to change, and the more necessary it is to... Amplify its rate of change during the monitoring process Indicates the characteristics of its data changes; when The smaller the value, the weaker the trend of change in the characteristic value of the power parameters above time point t in the time series direction. In other words, the change characteristic of the t-th time point is closer to the overall characteristic of power consumption (electricity usage). Therefore, by... Amplify β t The overall variation characteristics of the power parameters are represented and used as the parameter variation characteristic value at the t-th time point. Finally, the parameters representing the monotonic intervals are... After adding the adjusted data change characteristics (similar to a weighted average process), the result is obtained through d. t This represents the characteristic value of parameter change at the t-th time point.

[0074] Furthermore, after obtaining the characteristic values ​​of power parameter changes at each point, the monitoring disturbance degree of data in different dimensions is analyzed in conjunction with the above analysis.

[0075] In some embodiments, step S30, using parameter change feature values, determines the parameter detection perturbation degree in different dimensions, and may include the following:

[0076] First, obtain the first derivative time sequence of the current change curve, and then obtain the second derivative time sequence of the voltage change curve.

[0077] Among them, the first derivative time series is the derivative time series corresponding to the current change curve, and the second derivative time series is the derivative time series corresponding to the voltage change curve.

[0078] Next, the amount of monitoring data in each dimension is obtained, the average difference between the monitoring data of the target time point and the adjacent time points in each dimension is obtained, and the average difference between the parameter change characteristic values ​​of the target time point and the adjacent time points is obtained.

[0079] Because it contains multidimensional data, it is necessary to obtain the amount of monitoring data for each dimension, the corresponding mean difference, and the corresponding mean difference.

[0080] Specifically, taking the t-th time series point as the target time series point, the monitoring data volume under different dimensions is obtained, and the mean difference Δx between the monitoring data of the t-th time series point and the adjacent time series points under different dimensions is obtained. t And obtain the mean difference between the parameter change feature values ​​of the t-th time series point and its adjacent time series points under different dimensions, where the mean difference and the mean difference can be obtained after standardization.

[0081] Next, the parameter detection perturbation degree of each dimension is determined by using the first derivative time series, the second derivative time series, the amount of monitoring data, the mean difference, and the mean difference.

[0082] Specifically, the frequent data change parameters of the target time series point are determined by using the mean difference and the mean difference value. Then, the first calculation parameter of each dimension is determined by using the frequent data change parameters and the amount of monitoring data. The derivative time series coefficients of the voltage change curve and the current change curve are determined by using the first derivative time series and the second derivative time series. The parameter detection disturbance degree under each dimension is determined by using the preset exponential function, the first calculation parameter and the derivative time series coefficients.

[0083] The parameter detection perturbation degree is calculated as follows:

[0084]

[0085] in, To detect the perturbation degree of the parameters, tanh is the hyperbolic tangent function, I ′ U is the time series of the first derivative corresponding to the current change curve. ′ Let be the time series of the second derivative of the voltage change curve, r be the correlation calculation function, n be the amount of monitoring data in the current dimension, and Δx be the time series of the voltage change curve. t Δd represents the standardized mean difference between the monitoring data of the t-th time series point and its adjacent time series points in the current dimension of data. t This represents the mean difference between the parameter change characteristic values ​​of the t-th time series point and its adjacent time series points.

[0086] Based on the above analysis, electrical equipment in a power system does not frequently adjust its power parameters during use. To analyze the detection disturbance degree of a certain dimension of data, it is necessary to analyze the frequency of change of that dimension of data: Δx t The difference Δx between the t-th time series point in the current dimension and its neighboring time series points was calculated. t The larger the value, the higher the degree of fluctuation in the power parameters at the current time point. Furthermore, besides numerical fluctuations, the greater the difference in parameter change characteristic values ​​between the current time point and adjacent time points, the more frequent and obvious the data changes reflected in the power parameter fluctuations, meaning a greater degree of disturbance detected in the current parameters. Therefore, through Δx... t ×Δd t As a parameter representing the most frequent changes in data at time point t, after averaging it, we use... As one of the parameters for calculating the disturbance degree of power data in a certain dimension.

[0087] In this embodiment, electricity consumption data reflects the power system's outcome parameters, and the monitored data mainly refers to current and voltage data. For a power system, the changes in current and voltage data should be directly proportional; however, this proportionality can be affected when the monitoring process is disturbed. ′ U ′ The correlation coefficients of the time series derivatives of the voltage and current variation curves were calculated. Based on the above analysis, under normal circumstances, r(I) ′ U ′ The value of r(I) should be relatively large; however, the more interference the current data detection is subject to, the smaller the value of r(I) becomes. ′ U ′ The value of ) is relatively small. Therefore, through After adjusting the logical relationship and setting the value range to (0, 1), it is used as another parameter for calculating the detection perturbation degree.

[0088] Finally, after combining the two parameters, the range of values ​​is set between (0, 1) using tanh, and then... This indicates the disturbance degree of parameter detection for a certain dimension of power data. The larger the value, the stronger the interference experienced during the detection process of the current dimension of data.

[0089] Furthermore, after acquiring the power data disturbance degree from various dimensions, an anomaly detection algorithm for power instrument monitoring data is designed by adjusting the LOF anomaly detection algorithm. The LOF algorithm analyzes abnormal data by calculating the local density of data points in the data space, and the distance between data points is an important parameter directly reflecting the local density of data points. Traditional LOF algorithms mainly use Euclidean distance as a distance metric between data points, but for power monitoring data, this method ignores the temporal variation characteristics of power data, therefore, further optimization of the distance metric is needed.

[0090] As is known, step S40 uses parameters to detect perturbation and obtains the dimensional distance between at least two data points under different dimensions. The dimensional distance is used to determine the data monitoring result, and then the abnormal data is cleaned based on the monitoring result. This may include the following steps:

[0091] Obtain data values ​​for at least two test data points in different dimensions, and obtain parameter change feature values ​​for at least two test data points.

[0092] Here, data value refers to the data value of different data points under the current dimension, and the data points under test can also be time series points.

[0093] Specifically, the data values ​​of two different test data points in the current dimension are obtained, and the parameter change feature values ​​corresponding to the two different test data points are obtained respectively.

[0094] By using data values, parameter change characteristic values, and parameter detection perturbation, the dimensional distance between at least two test data points in different dimensions can be determined.

[0095] The parameter change characteristic value and the parameter detection perturbation degree can both be obtained from the aforementioned calculation process.

[0096] Specifically, taking data point a and data point b as an example, we obtain the first difference between the data value of the j-th dimension of data point a and the data value of the j-th dimension of data point b, and obtain the second difference between the parameter change feature value of data point a and the parameter change feature value of data point b. We then use the dimension data value, parameter detection perturbation degree, first difference and second difference to determine the dimensional distance between data point a and data point b in the data space.

[0097] The dimensional distance is calculated as follows:

[0098]

[0099] Where s(a,b) is the dimensional distance between multidimensional data points a and b in the data space, and also the distance between power data anomalies; Let 'a' be the data value of the j-th dimension of data point 'a'. Let d be the data value of the j-th dimension of data point b. a Let d be the characteristic value of parameter change for data point a. b Let b be the feature value of parameter variation for data point b, and m be the number of dimensions. The perturbation degree of the parameter in the j-th dimension is detected.

[0100] in, The distances between data points a and b across various dimensions were calculated, and adjusted using the detected perturbation level of that dimension. A higher detected perturbation level in that dimension resulted in a greater weight for the distance calculation between data points a and b in that dimension, thus highlighting the distance between potentially anomalous and normal data. Additionally, through (d... a -d b ) 2 The distance between data points a and b, considering the characteristic distance of power parameter changes, is used as another parameter for distance calculation. That is, in addition to the differences between actual measured values, the characteristic power parameter changes are also taken into account in the distance between data points.

[0101] In some embodiments, after obtaining the distance metric, the LOF algorithm also needs to specify the neighborhood range k. To further reduce the impact of different data variation characteristics on the anomaly detection process, the average number of data points contained in each monotonic interval of the voltage and current curves is used as the k parameter to improve the accuracy of anomaly detection, i.e., optimize the data monitoring results.

[0102] That is, the first average value of the data points contained in each monotonic interval of the current change curve and the second average value of the data points contained in each monotonic interval of the voltage change curve are obtained; the data monitoring results are optimized using the first and second average values; after obtaining the detection parameters, the current power instrument monitoring data is detected by the LOF algorithm to obtain abnormal data.

[0103] In some embodiments, abnormal data is identified using monitoring results, deleted, and then filled in according to the time sequence using interpolation. That is, after deleting the detected abnormal data, common interpolation methods are used to fill in the data according to the corresponding time sequence to achieve data cleaning.

[0104] In some embodiments, the data after cleaning is acquired and the data after cleaning of the power instrument is output.

[0105] In this application, a fitting curve for monitoring data is established to analyze the overall trend of data changes, providing parameters for the anomaly detection and judgment process. Then, the detection disturbance degree of the two-dimensional data is analyzed by combining the electricity consumption change trend and the characteristics of current and voltage curves. This further improves the accuracy of data change characteristic analysis while evaluating the interference characteristics of detection, providing parameters for subsequent adjustments to the anomaly detection process. Finally, the distance measurement method is optimized by combining the disturbance degree and data change characteristics of each dimension of data, and the anomaly detection algorithm is optimized based on the change characteristics of the fitting curve. Compared with existing technologies, the detection process adds analysis of electricity consumption data change characteristics, reduces the interference of power data changes on anomaly detection, improves detection accuracy, and further increases the accuracy of power system analysis.

[0106] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0107] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method of processing monitoring data of an electric power instrument, characterized by, The method comprises: acquiring monitoring data of the power instrument, and determining parameter change curves of different dimensions according to the monitoring data; determining power consumption trend characteristics of each time sequence point by using the parameter change curves of different dimensions, and further determining parameter change characteristic values of each time sequence point, comprising: determining a power consumption change curve by using the parameter change curves of different dimensions; wherein the parameter change curves comprise current change curves and voltage change curves; acquiring derivatives of each time sequence point on the power consumption change curve, and determining a derivative mean value of each time sequence point on the power consumption change curve by using the derivatives; determining the power consumption trend characteristics of each time sequence point by using a preset activation function, the derivatives and the derivative mean value; acquiring relative time sequence positions of each time sequence point in a preset interval in the parameter change curves of different dimensions, and further determining the parameter change characteristic values of each time sequence point by using the relative time sequence positions and the power consumption trend characteristics; determining parameter detection disturbance degrees in different dimensions by using the parameter change characteristic values, comprising: acquiring a first derivative time sequence of the current change curve, and acquiring a second derivative time sequence of the voltage change curve; acquiring a monitoring data amount in each dimension, acquiring a difference mean value of monitoring data of a target time sequence point and adjacent time sequence points in each dimension, and acquiring a difference value mean value of the parameter change characteristic values of the target time sequence point and the adjacent time sequence points; determining a data change frequent parameter of the target time sequence point by using the difference mean value and the difference value mean value, and further determining a first calculation parameter of each dimension by using the data change frequent parameter and the monitoring data amount; determining derivative time sequence sequence coefficients of the voltage change curve and the current change curve by using the first derivative time sequence and the second derivative time sequence; determining the parameter detection disturbance degrees in each dimension by using a preset exponential function, the first calculation parameter and the derivative time sequence sequence coefficients; acquiring dimension distances of at least two to-be-measured data points in different dimensions by using the parameter detection disturbance degrees, determining a data monitoring result by using the dimension distances, and further cleaning abnormal data according to the monitoring result, comprising: acquiring data values of the at least two to-be-measured data points in different dimensions, and acquiring the parameter change characteristic values of the at least two to-be-measured data points; determining the dimension distances of the at least two to-be-measured data points in different dimensions by using the data values, the parameter change characteristic values and the parameter detection disturbance degrees; determining the abnormal data by using the dimension distances in different dimensions, and further determining the data monitoring result.

2. The monitoring data processing method of a power instrument according to claim 1, characterized by, acquiring monitoring data of the power instrument, and determining parameter change curves of different dimensions according to the monitoring data, comprising: detecting the target node by using the power instrument to acquire the monitoring data, wherein the monitoring data comprises monitoring data of different dimensions; respectively performing curve fitting on the monitoring data of different dimensions to acquire the parameter change curves of different dimensions.

3. The monitoring data processing method of power instruments according to claim 1, characterized in that, acquiring relative time sequence positions of each time sequence point in a preset interval in the parameter change curves of different dimensions, and further determining the parameter change characteristic values of each time sequence point by using the relative time sequence positions and the power consumption trend characteristics, comprising: The first relative time position of each time point in the preset interval of the current change curve is obtained, and the second relative time position of each time point in the preset interval of the voltage change curve is obtained; The first derivative corresponding to each time point in the current change curve is obtained, and the second derivative corresponding to each time point in the voltage change curve is obtained, wherein the first derivative and the second derivative are obtained after standardization processing; The parameter change characteristic value of each time point is determined by using the first relative time position, the second relative time position, the first derivative, the second derivative and the power consumption trend characteristic.

4. The monitoring data processing method of power instruments according to claim 3, characterized in that, The parameter change characteristic value of each time point is determined by using the first relative time position, the second relative time position, the first derivative, the second derivative and the power consumption trend characteristic, including: The overall change characteristic of the power parameter is determined by using the first relative time position, the second relative time position and the power consumption trend characteristic; The data change characteristic is determined by using the first relative time position, the second relative time position, the first derivative and the second derivative; The parameter change characteristic value of each time point is determined by using the overall change characteristic of the power parameter and the data change characteristic.

5. The monitoring data processing method of power instruments according to claim 1, characterized in that, Further comprising: The first data mean of the data points contained in each monotonic interval in the current change curve is obtained, and the second data mean of the data points contained in each monotonic interval in the voltage change curve is obtained; The data monitoring result is optimized by using the first data mean and the second data mean.

6. The monitoring data processing method of power instruments according to claim 1, characterized in that, According to the monitoring result, the abnormal data is cleaned, including: The abnormal data is deleted in the monitoring result, and then the interpolation method is used to fill in the time sequence order.

Images