Line loss abnormity identification method and device of power network and electronic equipment
By extracting the fluctuating component data sequences in the power network and building multiple regression models, comparing the model differences, the problem of low accuracy in recognition of line loss anomalies caused by increasing volatility of wind and solar energy is solved, and higher recognition accuracy is achieved sex.
Patent Information
- Application Number
- CN202411059073.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-02
- Publication Date
- 2024-11-15
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing power network line loss abnormality recognition methods are difficult to accurately identify line loss abnormalities in the face of increasing wind energy and solar energy fluctuations, resulting in low accuracy of the identification results.
By obtaining the line loss data of the power network, extracting the fluctuation component data sequence, building multiple regression models, and comparing the difference between these models to determine the results of line loss abnormality identification, eliminating interference from new energy fluctuations, and improving identification accuracy .
Accurate identification of line loss abnormalities of power networks is achieved, the accuracy of identification results is improved, and the interference of new energy fluctuations on identification results is avoided.
Smart Images

Figure CN118965226A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of smart grids, and in particular to a method, device and electronic equipment for identifying abnormal line losses in a power network. Background Art
[0002] At present, the proportion of new energy sources connected to the grid is increasing. However, new energy sources, especially wind and solar energy, show significant volatility, which brings a series of complex challenges to the statistics and management of power network line losses. Line loss has a quadratic relationship with power output, and the fluctuations of new energy sources such as wind and solar energy are more intense, making it difficult for traditional statistical line loss abnormal data identification methods to cope with it. Abnormal data has caused line loss fluctuations and large negative losses in the system, exceeding the threshold of traditional fluctuations.
[0003] Although logical topology, statistical identification and other related technical means have been used to detect statistical line loss abnormal data, with the access of high proportions of new energy such as wind power and solar energy, the strong volatility has caused traditional methods to become increasingly limited, and it is difficult to accurately establish the relationship between statistical line loss abnormal data and theoretical line loss after the access of new energy. This means that although abnormal fluctuations can be detected, it is impossible to accurately evaluate their impact on actual line losses, which limits the in-depth understanding and processing of abnormal data, and then leads to low accuracy of line loss abnormality identification results for power networks.
[0004] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention
[0005] The embodiments of the present invention provide a method, device and electronic device for identifying line loss anomalies in an electric power network, so as to at least solve the technical problem that the accuracy of abnormal identification results is low in the related art of directly identifying line loss anomalies by comparing the differences between line loss data and reference data due to the drastic fluctuations of new energy sources such as wind energy and solar energy in the electric power network.
[0006] According to one aspect of an embodiment of the present invention, a method for identifying line loss anomalies in an electric power network is provided, comprising: acquiring line loss data of the electric power network; extracting a fluctuating component data sequence from the line loss data, wherein the fluctuating component data sequence includes data points in the line loss data whose fluctuation amplitude is greater than a predetermined amplitude; constructing N regression models based on the fluctuating component data sequence, wherein N is an integer greater than 2, and the N regression models are constructed based on data points that are not completely the same; determining the degree of difference between the N regression models; and determining a line loss anomaly identification result of the electric power network based on the degree of difference, wherein the line loss anomaly identification result is used to indicate whether there is a line loss anomaly in the electric power network.
[0007] According to another aspect of an embodiment of the present invention, there is provided a device for identifying line loss anomalies in an electric power network, comprising: a data acquisition module for acquiring line loss data of the electric power network; a sequence extraction module for extracting a fluctuating component data sequence in the line loss data, wherein the fluctuating component data sequence includes data points in the line loss data having a fluctuation amplitude greater than a predetermined amplitude; a model construction module for constructing N regression models based on the fluctuating component data sequence, wherein N is an integer greater than 2, and the N regression models are constructed based on data points that are not completely the same; a difference determination module for determining the difference between the N regression models; and an anomaly identification module for determining a line loss anomaly identification result of the electric power network based on the difference, wherein the line loss anomaly identification result is used to indicate whether there is a line loss anomaly in the electric power network.
[0008] According to another aspect of an embodiment of the present invention, there is also provided an electronic device, comprising one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement any one of the methods for identifying line loss anomalies in a power network.
[0009] In an embodiment of the present invention, the line loss data of the power network is obtained; a fluctuation component data sequence is extracted from the line loss data, wherein the fluctuation component data sequence includes data points in the line loss data whose fluctuation amplitude is greater than a predetermined amplitude; based on the fluctuation component data sequence, N regression models are constructed, wherein N is an integer greater than 2, and the N regression models are constructed based on data points that are not completely the same; the difference between the N regression models is determined; based on the difference, a line loss anomaly recognition result of the power network is determined, wherein the line loss anomaly recognition result is used to indicate whether there is a line loss anomaly in the power network, thereby achieving the goal of identifying a line loss anomaly based on line loss data. The method extracts a fluctuation component data sequence from the data, and constructs multiple regression models based on the data points in the fluctuation component data sequence. By comparing the differences between the multiple regression models, the purpose of accurately identifying line loss anomalies is achieved, thereby eliminating the interference of drastic fluctuations of new energy sources such as wind energy and solar energy in the power network on line loss anomaly identification and improving the accuracy of line loss anomaly identification. The method further solves the technical problem of low accuracy of anomaly identification results in the related technology of directly identifying line loss anomalies by comparing the differences between line loss data and reference data due to the drastic fluctuations of new energy sources such as wind energy and solar energy in the power network. BRIEF DESCRIPTION OF THE DRAWINGS
[0010] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:
[0011] Figure 1 is a flow chart of a method for identifying abnormal line loss in a power network according to an embodiment of the present invention;
[0012] Figure 2 is a flow chart of an optional method for identifying abnormal line loss in a power network according to an embodiment of the present invention;
[0013] Figure 3 is a schematic diagram of an optional error sequence according to an embodiment of the present invention;
[0014] Figure 4 is a schematic diagram of a device for identifying abnormal line loss in a power network according to an embodiment of the present invention. DETAILED DESCRIPTION
[0015] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.
[0016] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0017] First, to facilitate understanding of the embodiments of the present invention, some terms or nouns involved in the present invention are explained below:
[0018] Autoregressive Moving Average model with Differencing (ARMA), a time series prediction model that combines the features of an autoregressive model (AR) and a moving average model (MA). The ARMA model performs differential processing on the time series and then uses autoregressive and moving average terms to capture trends and periodic changes in the series to predict future values. According to an embodiment of the present invention, a method embodiment for identifying line loss anomalies in a power network is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in an order different from that herein.
[0019] Figure 1 is a flow chart of a method for identifying abnormal line loss in a power network according to an embodiment of the present invention. Figure 1 As shown, the method comprises the following steps:
[0020] Step S102, obtaining line loss data of the power network;
[0021] Optionally, the line loss data can be obtained from the power grid line loss management system. The line loss data of the power network refers to the power loss caused by line resistance, inductance and other reasons during the transmission process. During the power transmission process, electricity will be transmitted to the end user through the transmission line, but due to the resistance, inductance and other characteristics of the line itself, part of the electric energy will be converted into heat energy or other forms of energy and lost. This part of the lost electric energy is called line loss. Line loss data can be measured by, but not limited to, line loss rate. The line loss rate refers to the ratio of the power loss caused by resistance, inductance, capacitance and other reasons in the transmission line to the useful electric energy transmitted. By monitoring and calculating the line loss rate, the transmission efficiency and line quality of the power network can be evaluated, and line faults or large energy loss problems can be discovered in time.
[0022] Step S104, extracting a fluctuation component data sequence from the line loss data, wherein the fluctuation component data sequence includes data points in the line loss data whose fluctuation amplitude is greater than a predetermined amplitude;
[0023] Optionally, the line loss data may include data points corresponding to multiple different sampling moments, and the data points may be line loss rates collected at corresponding moments. The line loss data can be understood as the original line loss data collected in real time, and data points with fluctuation amplitudes greater than a predetermined amplitude are screened out from the original line loss data to form a new data sequence, namely, a fluctuation component data sequence. These data points usually represent volatility or abnormal conditions in the line loss data, that is, data points that may affect the stability and operation of the power network system. A fluctuation amplitude greater than a predetermined amplitude is used to indicate that the amplitude of the fluctuation in the line loss data exceeds a pre-set threshold or standard; the predetermined amplitude can be set according to actual conditions and needs, and is used to identify abnormal fluctuations in the line loss data.
[0024] In the above method, by extracting the fluctuation component data sequence, data points in the line loss data with fluctuation amplitude greater than the predetermined amplitude can be identified to help discover potential abnormal situations or problems. At the same time, the fluctuation component data sequence reflects the volatility of the line loss data, which can be used to analyze the stability and reliability of the power network system and help prevent line failures or other problems. In addition, by analyzing the fluctuation component data sequence, the operation strategy of the power system can be adjusted in time, the abnormal fluctuations in the line loss data can be reduced, and the operation efficiency of the power network can be improved while reducing the operation cost.
[0025] Step S106, constructing N regression models based on the fluctuation component data sequence, where N is an integer greater than 2;
[0026] Optionally, N regression models are constructed based on data points that are not completely the same. The fluctuation component data sequence may include N data points, and the N data points correspond to different regression models respectively. For any data point among the N data points, the regression model corresponding to the any data point may be constructed based on other data points except the any data point.
[0027] In an optional embodiment, N regression models are constructed based on the fluctuation component data sequence, including: obtaining any one regression model among the N regression models by the following method: selecting N-1 data points from the N data points included in the fluctuation component data sequence; constructing any one regression model based on the N-1 data points; obtaining N regression models by adopting the method of obtaining any one regression model.
[0028] Optionally, the N regression models can correspond to N data points one by one. For any data point among the N data points, when constructing the regression model corresponding to the data point, the data point can be eliminated from the N data points to obtain N-1 data points; based on the other N-1 data points except the data point, the regression model corresponding to the data point is constructed. By eliminating line loss data points one by one to construct the regression model, the influence of each data point on the overall model is fully considered, so that the subsequent determination of abnormal points is more global and the overall characteristics of the data are integrated.
[0029] In an optional embodiment, an arbitrary regression model is constructed based on N-1 data points, including: determining the difference order and the regression type based on the N-1 data points; constructing an initial regression model based on the difference order and the regression type; estimating regression parameters of the initial regression model to obtain a first parameter estimation result, wherein the first parameter estimation result at least includes a coefficient estimation result of an autoregressive term and a coefficient estimation result of a moving average term in the initial regression model; updating parameters of the initial regression model based on the first parameter estimation result to obtain a first regression model; optimizing initial order information in the first regression model to obtain a second regression model, wherein the initial order information includes an initial order of an autoregressive term and an initial order of a moving average term; updating regression parameters of the second regression model to obtain an arbitrary regression model; and obtaining N regression models in the manner of obtaining an arbitrary regression model.
[0030] Optionally, the initial regression model can be constructed based on an autoregressive moving average model with differencing (ARMA). The autoregressive moving average regression model can include three categories: autoregressive (AR) model, moving average regression (MA) model, and autoregressive moving average regression (ARMA) model. Before constructing the initial regression model, determining the difference order and regression type can help select a suitable model structure for constructing the initial regression model, so as to better capture the characteristics and laws of N-1 data points. By estimating and updating the parameters of the initial regression model, more accurate model parameters can be obtained, and the fit and prediction accuracy of the first regression model can be improved. By optimizing the order information, it can help select the appropriate order of the autoregressive term and the moving average term, avoid the problem of overfitting or underfitting, and improve the generalization ability of the constructed second regression model.
[0031] In the above method, by iteratively updating parameters and optimizing order information for multiple times, a more accurate and stable regression model can be obtained, thereby improving the prediction accuracy of the model; the use of automated methods for modeling and prediction reduces human subjective interference and improves the efficiency and accuracy of modeling; multiple regression models can be obtained as needed, so that the performance of different models can be compared, thereby reflecting the impact of each data point on the overall model, making the subsequent determination of abnormal points more global and integrating the overall characteristics of the data.
[0032] In an optional embodiment, the difference order and the regression type are determined based on N-1 data points, including: based on the N-1 data points, the difference order is determined in the following manner: performing multiple difference processing on the N-1 data points to obtain multiple difference results, wherein the multiple difference processing and the multiple difference results correspond one-to-one; determining the stability scores corresponding to the multiple difference results respectively; determining the difference order based on the number of differences corresponding to the difference result with the highest stability score among the multiple difference results.
[0033] It should be noted that by performing multiple differential processing on N-1 data points, and then evaluating the stability of each differential result, the most suitable differential order is finally determined. The differential order is an important parameter in the differential autoregressive sliding average regression model, which affects the prediction effect of the model. Selecting a suitable differential order can reduce the non-stationarity of the data, making the data easier to model and predict, thereby improving the accuracy and stability of the model. In the above method, by scoring the stability of multiple differential results, the differential order can be automatically determined without manual subjective intervention, simplifying the model selection process. Selecting the number of differentials corresponding to the differential result with the highest stability score can effectively reduce the non-stationarity of the data and improve the stability and reliability of the model. By reasonably selecting the differential order, overfitting of data can be avoided, the generalization ability of the model can be improved, and the situation of performing well on training data but poorly on unknown data can be avoided.
[0034] Optionally, the data points included in the fluctuation component data sequence can be visualized and analyzed to observe its volatility and trend. If there is an obvious trend or periodic change between the data points, some differential processing may be required. First, the N-1 data points included in the fluctuation component data sequence can be differentially processed, that is, the difference between adjacent data points is calculated. Then observe whether the processed data sequence tends to be stable. If the data sequence still has a trend or periodic fluctuation after the preliminary differential processing, further differential processing can be performed to calculate the difference between adjacent data points. Repeat the above operations until the processed fluctuation component data sequence tends to be stable. The order of differential processing required can be determined based on the observation of the stationarity of the data sequence after differential processing.
[0035] It should be noted that the difference autoregressive moving average regression model can include three categories: AR regression model, MA regression model, and ARMA regression model. In the regression process, it is necessary to determine the appropriate regression model type (i.e., regression type) to obtain the best fitting effect. The regression type can be determined based on the data characteristics of N-1 data points, and the autocorrelation coefficients shown in Table 1 are Partial autocorrelation coefficient The tailing and truncated nature of the data determines the regression type:
[0036] Table 1 Identification criteria for the difference autoregressive moving average model
[0037]
[0038] in, represents the autocorrelation coefficient, which is used to measure the correlation between a data series consisting of N-1 data points at different time points; represents the partial autocorrelation coefficient, which is used to measure the correlation between a specific lag term and the current value after removing the influence of other lag terms in the data series. Tailing means that the data series will never tend to zero; truncation means that the data series will quickly tend to zero after exceeding the k value. According to the data shown in Table 1 and The tailing and truncated nature of the data can determine the appropriate regression model type (AR, MA, ARMA) in the following way:
[0039] 1) If After a certain order, it decreases rapidly and truncates, while If the regression type decreases rapidly and is truncated after a certain order, then it is determined to be ARMA.
[0040] 2) If After a certain order, it decreases rapidly and truncates, while If it is approximately zero after a certain order, the regression type is determined to be AR.
[0041] 3) If After a certain order, it decreases rapidly and truncates, while If it drops rapidly and truncates after a certain order, the regression type is determined to be MA.
[0042] In an optional embodiment, regression parameters of the initial regression model are estimated to obtain a first parameter estimation result, including: estimating the regression coefficient of the initial regression model to obtain an initial parameter estimation result; constructing a likelihood function based on the initial parameter estimation result and N-1 data points; optimizing the initial parameter estimation result with the goal of maximizing the function value of the likelihood function to obtain a first parameter estimation result.
[0043] Optionally, by estimating the regression coefficients of the initial regression model, an initial parameter estimation result can be obtained for constructing a likelihood function. Based on the initial parameter estimation result and N-1 data points, a likelihood function is constructed to evaluate the degree of fit and prediction accuracy of the model. With the goal of maximizing the function value of the likelihood function, the initial parameter estimation result is optimized to obtain a first parameter estimation result, so that the model fits the data better and the accuracy of parameter estimation is improved.
[0044] In the above method, by optimizing the parameter estimation results, the model can better fit the data and improve the prediction accuracy and stability of the model. Based on the maximization of the likelihood function, the parameters of the model can be automatically adjusted to make the model more consistent with the actual situation of the data and reduce the subjectivity of human intervention. By optimizing the initial parameter estimation results, the efficiency of parameter estimation can be improved, the time for establishing and optimizing the first parameter estimation results can be shortened, and work efficiency can be improved.
[0045] Optionally, the initial regression model is constructed as a differential autoregressive moving average model, and the regression coefficient of the differential autoregressive moving average model is estimated using the maximum likelihood method. The differential autoregressive moving average model can be expressed as: (1-φ1B-φ2B 2 -…-φpB p )(1-B) d Xt=(1+θ1B+θ2B p +…+θqB q )εt; where p represents the order of the autoregressive term, d represents the number of differences (i.e., the difference order), and q represents the order of the moving average term. φ1, φ2, …, φp represent the coefficients of the autoregressive term, θ1, θ2, …, θq represent the coefficients of the moving average term, B represents the lag operator, Xt represents the time series data (i.e., the original data of the time series), and εt represents the white noise error.
[0046] The coefficients in the difference autoregressive moving average model are estimated by maximizing the likelihood function. Assuming that εt is an independent and identically distributed normal distribution, the likelihood function can be expressed as follows: L(φ1,φ2,…,φp,θ1,θ2,…,θq|X1,X2,…,Xn).
[0047] In an optional embodiment, the initial order information in the first regression model is optimized to obtain the second regression model, including: based on the first regression model, determining the first prediction points corresponding to N-1 data points respectively; based on the N-1 data points and the first prediction points corresponding to the N-1 data points respectively, obtaining the sum of square errors corresponding to the first regression model, wherein the sum of square errors is used to indicate the degree of difference between the N-1 data points and the corresponding first prediction points; based on the sum of square errors and the initial order information, constructing a Bayesian information criterion function, wherein the Bayesian information criterion function is used to indicate the relationship between the sum of square errors, the initial order information and the initial order information; with the minimum function value of the Bayesian information criterion function as the goal, optimizing the initial order information to obtain the target order information; based on the target order information, updating the initial order information in the first regression model to obtain the second regression model.
[0048] It can be understood that, first, the N-1 data points are fitted by the first regression model to obtain the first prediction point corresponding to each data point. Then, the sum of squared errors between each data point and the corresponding first prediction point is calculated to measure the accuracy of the model fitting. Then, a Bayesian information criterion function is constructed based on the sum of squared errors and the initial order information. The Bayesian information criterion function helps to select the most appropriate order information under the consideration of model complexity and fitting accuracy. By optimizing the function value of the Bayesian information criterion function, the target order information can be obtained, that is, the model with the simplest structure is selected under the premise of ensuring the accuracy of the model. Finally, the initial order information in the first regression model is updated according to the target order information to obtain the second regression model. In the above manner, determining the first prediction point corresponding to each data point can better understand the degree of fit of the model to the data. Calculating the sum of squared errors can quantify the prediction accuracy of the model and help evaluate the effect of the model. Constructing the Bayesian information criterion function can comprehensively consider the balance between the sum of squared errors and model complexity to avoid overfitting or underfitting. By optimizing the target order information, a more suitable second regression model can be obtained.
[0049] It should be noted that in the above method, by conducting multi-faceted analysis and evaluation of the data points and the first regression model, the accuracy and stability of the second regression model can be improved. By optimizing the Bayesian information criterion function, the obtained second regression model can be prevented from being too simple or complex, making the obtained second regression model more generalizable. Updating the order information of the model can make the obtained second regression model closer to the actual situation of the data and improve the predictive ability and interpretability of the second regression model. The above method comprehensively considers the predictive performance and complexity of the model, and can obtain a more reasonable and effective second regression model.
[0050] Optionally, based on N-1 data points and the first prediction points corresponding to the N-1 data points, an error sequence corresponding to the first regression model is obtained, and the sum of squares of errors corresponding to the data included in the error sequence corresponding to the first regression model is calculated. The degree of difference between each data point and the corresponding prediction point can be measured by the sum of squares of errors. Specifically, the first parameter estimation result obtained by estimating the regression parameters of the initial regression model can be used to fit the model with the N-1 data points to obtain the prediction value of the first regression model for the N-1 data points, that is, to obtain the first prediction points corresponding to the N-1 data points; the corresponding first prediction values are subtracted from the N-1 data points to obtain the error sequence corresponding to the first regression model; each error value in the error sequence corresponding to the first regression model is squared, and then all squared error values are added to obtain the model error sum of squares.
[0051] Optionally, according to the Bayesian Information Criterion, the order p of the autoregressive term and the order q of the moving average term of the differential autoregressive moving average regression model are optimized with the goal of minimizing the Bayesian Information Criterion (BIC) value of the data. The Bayesian Information Criterion function of the centralized and decentralized regression models can be expressed in the following form:
[0052]
[0053] Wherein, BIC represents the Bayesian Information Criterion value, and n=N-1.
[0054] The p and q orders that minimize the Bayesian information criterion function are considered to be optimal. By optimizing using the Bayesian information criterion, the p and q order table of the fluctuation component data can be obtained, as shown in Table 2. Based on Table 2, the order of the fluctuation component data when the BIC value is the smallest is determined as the difference order.
[0055] Table 2p, q order table
[0056] BIC value p=1 p=2 p=3 p=4 p=5 q=1 -11.4328 -11.5405 -11.5655 -11.5715 -11.5765 q=2 -11.4515 -11.5568 -11.5727 -11.5704 -11.5676 q=3 -11.5542 -11.5535 -11.5726 -11.5767 -11.7332 q=4 -11.5508 -11.5763 -11.5834 -11.5732 -11.7367 q=5 -11.5785 -11.5783 -11.5809 -11.5775 -11.5505
[0057] In an optional embodiment, the regression parameters of the second regression model are updated to obtain any regression model, including: based on the first regression model, determining that N-1 data points correspond to the second prediction points respectively; based on the N-1 data points, and the N-1 data points correspond to the second prediction points respectively, obtaining the error sequence corresponding to the second regression model; with the error sequence obeying the normal distribution as the goal, the regression parameters of the second regression model are updated to obtain any regression model.
[0058] Optionally, according to the first regression model, N-1 data points are used to predict the values of the corresponding data points respectively, and the second prediction points corresponding to the N-1 data points are obtained. Using these N-1 data points and the corresponding second prediction points, a second regression model is established, and the error sequence is obtained by calculating the error between the predicted value (i.e., the second prediction point) and the actual value (i.e., the data point). The error sequence of the obtained second regression model can be used to evaluate the fitting effect and prediction accuracy of the model. The smaller the regression error, the better the regression effect. In addition, if the model can accurately describe the correlation information of the original data sequence, the error sequence should present white noise characteristics. If the residuals are still correlated, it means that the model regression effect is not good and there may be unextracted information. The Shapiro-Wilk test method can be used to perform a white noise test to determine whether the mean and variance of the fitted residual sequence obey the normal distribution (white noise), thereby verifying the applicability of the differential autoregressive moving average regression characteristics in the statistical line loss fluctuation component data. By analyzing the error sequence, it can be determined whether the error obeys the normal distribution, and then the applicability and accuracy of the regression model can be determined. By updating the regression parameters of the second regression model, the fitting effect of the model can be improved, and any regression model with higher prediction accuracy and better performance can be obtained.
[0059] It should be noted that in the above method, by determining that N-1 data points correspond to the second prediction point, the prediction ability and accuracy of the model can be verified, thereby improving the reliability of the model. The error sequence corresponding to the second regression model can be obtained to evaluate the fitting effect and prediction accuracy of the model, providing a basis for the improvement of the model. By analyzing the error sequence, it is determined whether the model conforms to the normal distribution assumption and the parameters are updated, thereby determining any regression model with higher applicability and reliability, so that the model is more in line with the characteristics of the actual data and the prediction accuracy of the model is improved.
[0060] Step S108, determining the difference between the N regression models;
[0061] It should be noted that by setting each data point in the fluctuation component data sequence to correspond to a regression curve, if a data point is determined to be an abnormal data point, the regression curve corresponding to it will be significantly different from the regression curves corresponding to other data points. Therefore, by comparing the difference between the regression curves corresponding to the data points, it is possible to accurately determine whether the data point is abnormal.
[0062] In an optional embodiment, determining the difference between N regression models includes: determining the root mean square error and / or the mean absolute error between the N regression models; and determining the difference between the N regression models based on the root mean square error and / or the mean absolute error between the N regression models.
[0063] Optionally, the root mean square error (RMSE) and / or mean absolute error (MAE) of each regression model on a given data set are calculated. These error metrics reflect the prediction accuracy and fit of each regression model. By comparing the root mean square error and / or mean absolute error of each regression model, the difference between them can be evaluated. The smaller the error value, the better the prediction effect of the model, so the difference can be compared and judged by the size of the error value.
[0064] It should be noted that the root mean square error and mean absolute error are commonly used regression model evaluation indicators, which can objectively measure the degree of fit and prediction accuracy of the model. By comparing the error metrics between different regression models, it is possible to determine which data points have a large difference between the regression models corresponding to other data points, which helps to accurately identify abnormal data points.
[0065] Step S110: determining a line loss anomaly identification result of the power network based on the difference, wherein the line loss anomaly identification result is used to indicate whether there is a line loss anomaly in the power network.
[0066] Optionally, the difference can be used to measure the degree of difference between N regression models, thereby reflecting the change between the regression curve corresponding to each fluctuation component data point and the regression curve corresponding to other data points. The degree of difference is used as a feature of the abnormal data of the statistical line loss fluctuation component, providing a quantitative basis for abnormality detection.
[0067] It should be noted that in the related art, the difference between the line loss data and the reference data is directly used to determine whether there is a line loss anomaly in the power network. However, due to the drastic fluctuations in new energy sources such as wind energy and solar energy in the power network, the accuracy of the abnormal identification result determined by directly comparing the difference between the line loss data and the reference data is low. Based on this, by constructing regression models corresponding to N data points, if a data point is determined to be an abnormal data point, the regression curve corresponding to it will be significantly different from the regression curves corresponding to other data points. By eliminating the line loss data points one by one and calculating the degree of difference between the regression models, the influence of each data point on the overall model is fully considered, making the determination of abnormal points more global, integrating the overall characteristics of the data, and effectively avoiding the interference of abnormal identification due to the drastic fluctuations in new energy sources such as wind energy and solar energy in the power network.
[0068] Through the above steps S102 to S110, it is possible to achieve the goal of constructing multiple regression models based on the fluctuation component data sequence extracted from the line loss data and based on the data points in the fluctuation component data sequence, and accurately identify the line loss anomaly by comparing the differences between the multiple regression models, thereby eliminating the interference of the drastic fluctuations of new energy sources such as wind energy and solar energy in the power network on the line loss anomaly identification and improving the technical effect of the accuracy of line loss anomaly identification, thereby solving the technical problem of low accuracy of abnormal identification results in the related technology of directly identifying line loss anomalies by comparing the differences between line loss data and reference data due to the relatively drastic fluctuations of new energy sources such as wind energy and solar energy in the power network.
[0069] Based on the above embodiments and optional embodiments, the present invention proposes an optional implementation mode: Figure 2 is a flow chart of an optional method for identifying abnormal line loss in a power network according to an embodiment of the present invention. Figure 2 As shown, each statistical line loss fluctuation component data point corresponds to a regression curve. If a data point is determined to be an abnormal data point, the regression curve corresponding to it will be significantly different from the regression curves corresponding to other data points. Therefore, it is possible to determine whether the data point is abnormal by comparing the degree of difference between the regression curves corresponding to the data point. The method includes:
[0070] S1, obtaining a fluctuation component data sequence: obtaining line loss data from a power grid line loss management system, processing the line loss data, and extracting a fluctuation component data sequence.
[0071] S2, using the difference autoregressive moving average regression model to regress the volatility component data series, including the following steps:
[0072] S21, determination of the difference order: ensuring the smoothness of the input difference autoregressive sliding average regression model data by difference calculation. When the differential data reaches a smooth state, the order of the corresponding difference autoregressive sliding average model is determined.
[0073] S22, regression type determination: The differential autoregressive moving average regression model can include three types: AR regression model, MA regression model, and ARMA regression model. In the regression process, it is necessary to determine the appropriate regression model type to obtain the best fitting effect. The autocorrelation coefficients shown in Table 1 above can be used to determine the appropriate regression model type. Partial autocorrelation coefficient The tailing and truncation properties of the data determine the regression type, where tailing means that the data series will never tend to zero; truncation means that the data series will quickly tend to zero after exceeding the k value.
[0074] S23, Regression parameter estimation: Estimate the regression coefficients of the difference autoregressive moving average model using the maximum likelihood method. The coefficients in the difference autoregressive moving average model are estimated by maximizing the likelihood function.
[0075] S24, model order optimization: According to the Bayesian information criterion, with the goal of minimizing the value of the Bayesian information criterion of the data, optimize the order p of the autoregressive term and the order q of the moving average term of the differential autoregressive moving average regression model.
[0076] The p and q orders that minimize the Bayesian information criterion function are considered to be optimal. By optimizing using the Bayesian information criterion, the p and q order tables of the fluctuation component data can be obtained as the difference order.
[0077] S25, regression parameter update: after the p and q orders are updated, the model parameters are re-estimated. Figure 3 is a schematic diagram of an optional error sequence according to an embodiment of the present invention, such as Figure 3 As shown in the figure, the smaller the regression error is, the better the regression effect is. In addition, if the model can accurately describe the correlation information of the original data sequence, the error sequence should show white noise characteristics. If the residuals are still correlated, it means that the model regression effect is not good and there may be unextracted information. The Shapiro-Wilk test method is used to perform a white noise test to determine whether the fitted residual sequence conforms to the mean and whether the variance obeys the normal distribution (white noise), thereby verifying the applicability of the differential autoregressive moving average regression characteristics in the statistical line loss fluctuation component data.
[0078] S3, iteratively remove the line loss data points one by one: remove one data point from the complete line loss data set (i.e., N data points) to form a new fluctuation component data series containing N-1 data points. For this new series, perform the regression operation in step S2 to build N-1 regression models, each of which excludes one data point from the original data set. After traversing all the data, N regression models are obtained.
[0079] S4, calculate the degree of change of the regression curve: calculate the degree of difference between N regression models (using statistical indicators such as RMSE and MAE). The degree of difference reflects the change between the regression curve corresponding to each fluctuation component data point and the regression curve corresponding to other data points.
[0080] S5, abnormal data feature extraction: The degree of difference is used as the feature of the abnormal data of the statistical line loss fluctuation component, which provides a quantitative basis for anomaly detection.
[0081] This embodiment can achieve at least one of the following effects: 1) This embodiment takes into account the data characteristics under different wind power and photovoltaic access ratios, can more flexibly adapt to changes in the grid structure, and is not easily affected by changes in the proportion of new energy access. A method for extracting abnormal statistical data of line loss fluctuation components with a high proportion of new energy access to the grid is proposed, the statistical data law of line loss fluctuation components is studied, and the big data mining technology is used to summarize the pattern of feature extraction of abnormal data of fluctuation components, providing a quantitative basis for abnormal detection. 2) The use of automated modeling and differential autoregressive moving average model reduces the dependence on manual intervention and improves the processing efficiency of large-scale data. Accurate regression of fluctuation component data through differential autoregressive moving average model not only improves the detection accuracy of abnormal data, but also filters noise, making the detection of abnormal points more robust and reliable. 3) By calculating the degree of difference between regression models, the determination of abnormal points is more objective and operational, which helps system operation and maintenance personnel to better understand and process abnormal line loss data. At the same time, by eliminating line loss data points one by one and calculating the degree of difference between regression models, the influence of each data point on the overall model is fully considered, making the determination of abnormal points more global and integrating the overall characteristics of the data. 4) The Bayesian information criterion is used to optimize the order of the differential autoregressive moving average model, which has better applicability to nonlinear models and complex models and avoids overfitting.
[0082] In this embodiment, a line loss anomaly identification device for a power network is also provided, and the device is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the terms "module" and "device" can implement a combination of software and / or hardware for a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.
[0083] According to an embodiment of the present invention, there is also provided an embodiment of a device for implementing the above-mentioned method for identifying abnormal line loss in a power network. Figure 4 is a schematic diagram of the structure of a device for identifying abnormal line loss in a power network according to an embodiment of the present invention. Figure 4 As shown, the above-mentioned line loss anomaly identification device for the power network includes: a data acquisition module 400, a sequence extraction module 402, a model construction module 404, a difference determination module 406, and an anomaly identification module 408, wherein:
[0084] The data acquisition module 400 is used to acquire line loss data of the power network;
[0085] A sequence extraction module 402, connected to the data acquisition module 400, is used to extract a fluctuation component data sequence in the line loss data, wherein the fluctuation component data sequence includes data points in the line loss data whose fluctuation amplitude is greater than a predetermined amplitude;
[0086] A model building module 404, connected to the sequence extraction module 402, is used to build N regression models based on the fluctuation component data sequence, wherein N is an integer greater than 2, and the N regression models are built based on data points that are not completely the same;
[0087] A difference determination module 406, connected to the model building module 404, is used to determine the difference between the N regression models;
[0088] The abnormality identification module 408 is connected to the difference determination module 406 and is used to determine the line loss abnormality identification result of the power network based on the difference, wherein the line loss abnormality identification result is used to indicate whether there is a line loss abnormality in the power network.
[0089] In the embodiment of the present invention, a data acquisition module 400 is provided to acquire line loss data of the power network; a sequence extraction module 402 is connected to the data acquisition module 400 and is used to extract a fluctuation component data sequence in the line loss data, wherein the fluctuation component data sequence includes data points in the line loss data whose fluctuation amplitude is greater than a predetermined amplitude; a model construction module 404 is connected to the sequence extraction module 402 and is used to construct N regression models based on the fluctuation component data sequence, wherein N is an integer greater than 2, and the N regression models are constructed based on data points that are not completely the same; a difference determination module 406 is connected to the model construction module 404 and is used to determine the difference between the N regression models; an abnormality identification module 408 is connected to the difference determination module 406 and is used to determine the power network based on the difference. The line loss anomaly identification result of the network is obtained, wherein the line loss anomaly identification result is used to indicate whether there is a line loss anomaly in the power network, and a plurality of regression models are constructed based on the fluctuation component data sequence extracted from the line loss data and the data points in the fluctuation component data sequence. The purpose of accurately identifying the line loss anomaly is achieved by comparing the difference between the plurality of regression models, thereby eliminating the interference of the drastic fluctuation of new energy sources such as wind energy and solar energy in the power network on the line loss anomaly identification and improving the technical effect of the accuracy of the line loss anomaly identification, and further solving the technical problem of low accuracy of the anomaly identification result in the related technology of directly identifying the line loss anomaly by comparing the difference between the line loss data and the reference data due to the relatively drastic fluctuation of new energy sources such as wind energy and solar energy in the power network.
[0090] It should be noted that the above modules can be implemented by software or hardware. For example, for the latter, it can be implemented in the following ways: the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.
[0091] It should be noted that the above data acquisition module 400, sequence extraction module 402, model construction module 404, difference determination module 406, and abnormality identification module 408 correspond to steps S102 to S110 in the embodiment, and the examples and application scenarios implemented by the above modules and the corresponding steps are the same, but are not limited to the contents disclosed in the above embodiments. It should be noted that the above modules can be run in a computer terminal as part of the device.
[0092] It should be noted that the optional or preferred implementation of this embodiment can refer to the relevant description in the embodiment, which will not be repeated here.
[0093] The above-mentioned line loss anomaly identification device for the power network may also include a processor and a memory. The above-mentioned data acquisition module 400, sequence extraction module 402, model construction module 404, difference determination module 406, anomaly identification module 408, etc. are all stored in the memory as program modules, and the processor executes the above-mentioned program modules stored in the memory to realize the corresponding functions.
[0094] The processor includes a kernel, which retrieves the corresponding program module from the memory. The kernel may be one or more. The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. The memory includes at least one memory chip.
[0095] According to an embodiment of the present application, an embodiment of a non-volatile storage medium is also provided. Optionally, in this embodiment, the non-volatile storage medium includes a stored program, wherein when the program is executed, the device where the non-volatile storage medium is located is controlled to execute any of the above-mentioned methods for identifying abnormal line loss in a power network.
[0096] Optionally, in this embodiment, the non-volatile storage medium may be located in any computer terminal in a computer terminal group in a computer network, or in any mobile terminal in a mobile terminal group, and the non-volatile storage medium includes a stored program.
[0097] Optionally, when the program is running, the device where the non-volatile storage medium is located is controlled to perform the following functions: obtain line loss data of the power network; extract a fluctuating component data sequence from the line loss data, wherein the fluctuating component data sequence includes data points in the line loss data whose fluctuation amplitude is greater than a predetermined amplitude; based on the fluctuating component data sequence, construct N regression models, wherein N is an integer greater than 2, and the N regression models are constructed based on data points that are not completely the same; determine the difference between the N regression models; based on the difference, determine a line loss anomaly identification result of the power network, wherein the line loss anomaly identification result is used to indicate whether there is a line loss anomaly in the power network.
[0098] According to an embodiment of the present application, an embodiment of a processor is also provided. Optionally, in this embodiment, the processor is used to run a program, wherein when the program is run, any of the above-mentioned methods for identifying abnormal line loss in a power network is executed.
[0099] According to an embodiment of the present application, an embodiment of a computer program product is also provided. When executed on a data processing device, the program is suitable for executing the steps of initializing any one of the above-mentioned line loss anomaly identification methods for a power network.
[0100] Optionally, the above-mentioned computer program product, when executed on a data processing device, is suitable for executing a program initialized with the following method steps: obtaining line loss data of the power network; extracting a fluctuating component data sequence from the line loss data, wherein the fluctuating component data sequence includes data points in the line loss data whose fluctuation amplitude is greater than a predetermined amplitude; based on the fluctuating component data sequence, constructing N regression models, wherein N is an integer greater than 2, and the N regression models are constructed based on data points that are not completely the same; determining the difference between the N regression models; and based on the difference, determining a line loss anomaly identification result of the power network, wherein the line loss anomaly identification result is used to indicate whether there is a line loss anomaly in the power network.
[0101] An embodiment of the present invention provides an electronic device, which includes a processor, a memory, and a program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the program: acquiring line loss data of an electric power network; extracting a fluctuation component data sequence from the line loss data, wherein the fluctuation component data sequence includes data points in the line loss data whose fluctuation amplitude is greater than a predetermined amplitude; constructing N regression models based on the fluctuation component data sequence, wherein N is an integer greater than 2, and the N regression models are constructed based on data points that are not completely the same; determining the difference between the N regression models; and determining a line loss anomaly identification result of the electric power network based on the difference, wherein the line loss anomaly identification result is used to indicate whether there is a line loss anomaly in the electric power network.
[0102] The above sequence of the embodiments of the present invention is for description only and does not represent the superiority or inferiority of the embodiments.
[0103] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.
[0104] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the above modules can be a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, modules or indirect coupling or communication connection of modules, which can be electrical or other forms.
[0105] The modules described above as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed on multiple modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment.
[0106] In addition, each functional module in each embodiment of the present invention may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of software functional modules.
[0107] If the above-mentioned integrated module is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable non-volatile storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a non-volatile storage medium, including a number of instructions to enable a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the various embodiments of the present invention. The aforementioned non-volatile storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.
[0108] The above are only preferred embodiments of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. A method for identifying abnormal line loss in a power network, characterized in that: include: Obtain line loss data of the power network; Extracting a fluctuation component data sequence from the line loss data, wherein the fluctuation component data sequence includes data points in the line loss data whose fluctuation amplitude is greater than a predetermined amplitude; Based on the fluctuation component data sequence, construct N regression models, wherein N is an integer greater than 2, and the N regression models are constructed based on data points that are not completely the same; Determining the differences between the N regression models; Based on the difference, a line loss anomaly identification result of the power network is determined, wherein the line loss anomaly identification result is used to indicate whether there is a line loss anomaly in the power network.
2. The method according to claim 1, characterized in that The step of constructing N regression models based on the fluctuation component data sequence includes: Any one of the N regression models is obtained by: Selecting N-1 data points from the N data points included in the fluctuation component data sequence; Based on the N-1 data points, construct any one regression model; The N regression models are obtained by adopting the method of obtaining any one of the regression models.
3. The method according to claim 2, characterized in that The constructing any one of the regression models based on the N-1 data points includes: Based on the N-1 data points, determine the difference order and regression type; Based on the difference order and the regression type, construct an initial regression model; Performing regression parameter estimation on the initial regression model to obtain a first parameter estimation result, wherein the first parameter estimation result at least includes a coefficient estimation result of an autoregressive term and a coefficient estimation result of a moving average term in the initial regression model; Based on the first parameter estimation result, the parameters of the initial regression model are updated to obtain a first regression model; Optimizing the initial order information in the first regression model to obtain a second regression model, wherein the initial order information includes the initial order of the autoregressive term and the initial order of the moving average term; updating the regression parameters of the second regression model to obtain any one of the regression models; The N regression models are obtained by adopting the method of obtaining any one of the regression models.
4. The method according to claim 3, characterized in that Determining the difference order and regression type based on the N-1 data points includes: Based on the N-1 data points, the difference order is determined as follows: Performing multiple differential processing on the N-1 data points to obtain multiple differential results, wherein the multiple differential processing and the multiple differential results correspond one to one; Determine the stability scores corresponding to the plurality of differential results respectively; The difference order is determined based on the number of differences corresponding to the difference result with the highest stability score among the multiple difference results.
5. The method according to claim 3, characterized in that: The step of estimating the regression parameters of the initial regression model to obtain a first parameter estimation result includes: estimating the regression coefficient of the initial regression model to obtain an initial parameter estimation result; Constructing a likelihood function based on the initial parameter estimation result and the N-1 data points; The initial parameter estimation result is optimized with the goal of maximizing the function value of the likelihood function to obtain the first parameter estimation result.
6. The method according to claim 3, characterized in that: The step of optimizing the initial order information in the first regression model to obtain a second regression model includes: Based on the first regression model, determine first prediction points corresponding to the N-1 data points respectively; Based on the N-1 data points and the first prediction points respectively corresponding to the N-1 data points, a sum of square errors corresponding to the first regression model is obtained, wherein the sum of square errors is used to indicate the degree of difference between the N-1 data points and the corresponding first prediction points; Based on the error sum of squares and the initial order information, constructing a Bayesian information criterion function, wherein the Bayesian information criterion function is used to indicate the relationship between the error sum of squares, the initial order information and the initial order information; Taking the minimum function value of the Bayesian information criterion function as the goal, optimizing the initial order information to obtain target order information; Based on the target order information, the initial order information in the first regression model is updated to obtain the second regression model.
7. The method according to claim 3, characterized in that The updating of the regression parameters of the second regression model to obtain any one of the regression models comprises: Based on the first regression model, determining that the N-1 data points respectively correspond to second prediction points; Based on the N-1 data points, and the N-1 data points respectively corresponding to the second prediction points, an error sequence corresponding to the second regression model is obtained; With the goal of making the error sequence obey a normal distribution, the regression parameters of the second regression model are updated to obtain the arbitrary regression model.
8. The method according to any one of claims 1 to 7, characterized in that The determining the difference between the N regression models includes: Determining a root mean square error and / or a mean absolute error between the N regression models; The difference between the N regression models is determined based on a root mean square error and / or a mean absolute error between the N regression models.
9. A line loss anomaly identification device for a power network, characterized in that: include: A data acquisition module, used to acquire line loss data of the power network; A sequence extraction module, used to extract a fluctuation component data sequence from the line loss data, wherein the fluctuation component data sequence includes data points in the line loss data whose fluctuation amplitude is greater than a predetermined amplitude; A model building module is used to build N regression models based on the fluctuation component data sequence, wherein: N is an integer greater than 2, and the N regression models are constructed based on data points that are not completely the same; A difference determination module, used to determine the difference between the N regression models; The abnormality identification module is used to determine a line loss abnormality identification result of the power network based on the difference, wherein the line loss abnormality identification result is used to indicate whether there is a line loss abnormality in the power network.
10. An electronic device, characterized in that: It includes one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the line loss anomaly identification method for the power network as described in any one of claims 1 to 8.
Citation Information
Patent Citations
Electricity utilization topology identification system for user-transformer relation in transformer area
CN114048611A
New energy access power grid line loss abnormal data identification method and device
CN115758197A
Line loss abnormity determination method and device
CN115764872A
Method for analyzing correlation between different line loss actions
US20200366092A1
Line loss detection load balancing method and apparatus used for gateway and gateway device
WO2023164895A1
