Line loss abnormality detection method, device, computer equipment and storage medium

Abstract

The present application relates to a line loss anomaly detection method, device, computer equipment, storage medium and computer program product. The method comprises: performing wavelet transform processing on the line loss value at each historical moment in the line loss value database to obtain the high-frequency component and low-frequency component corresponding to the line loss value at each historical moment; determining the high-frequency component prediction value at a preset moment in the current time period according to the high-frequency component corresponding to the line loss value at each historical moment, and determining the low-frequency component prediction value at a preset moment according to the low-frequency component corresponding to the line loss value at each historical moment; determining the predicted line loss value at a preset moment according to the high-frequency component prediction value at the preset moment and the low-frequency component prediction value at the preset moment; determining whether the real line loss value at the preset moment is abnormal according to the predicted line loss value at the preset moment and the real line loss value at the preset moment. Thus, the labor cost is reduced, the efficiency of line loss anomaly analysis is improved, and the accuracy of the line loss anomaly analysis results is improved.

Description

Technical Field

[0001] The present application relates to the technical field of electric power grids, and in particular to a line loss anomaly detection method, apparatus, computer equipment, storage medium and computer program product. Background Art

[0002] Line loss is an important indicator for evaluating the operating performance of the power transmission network. In the related technology, the line loss anomaly analysis method of the power transmission and transformation equipment is mainly: after manually screening out the high-loss and negative-loss equipment, the power supply side and the supply side are compared and analyzed, and the line loss of the equipment connected to the previous voltage level or the same voltage level is combined to determine the anomaly. The above-mentioned line loss anomaly analysis method has the following problems: First, the manual analysis process is more affected by the outside world and there are many uncontrollable factors, resulting in low accuracy of the line loss anomaly analysis results; second, a large amount of manpower costs are required; third, the efficiency is low. Therefore, there is an urgent need for a line loss anomaly detection method to solve the above problems. Summary of the invention

[0003] Based on this, it is necessary to provide a line loss anomaly detection method, device, computer equipment, computer readable storage medium and computer program product that can improve the accuracy of line loss anomaly analysis results in response to the above technical problems.

[0004] In a first aspect, the present application provides a line loss anomaly detection method. The method comprises:

[0005] Establishing a line loss value database; the line loss value database contains the line loss value at each historical moment in a preset historical time period;

[0006] Perform wavelet transform processing on the line loss value at each historical moment in the line loss value database to obtain the high-frequency component and low-frequency component corresponding to the line loss value at each historical moment;

[0007] According to the high-frequency component corresponding to the line loss value at each historical moment, the predicted value of the high-frequency component at the preset moment in the current time period is determined; according to the low-frequency component corresponding to the line loss value at each historical moment, the predicted value of the low-frequency component at the preset moment is determined, and the current time period is determined by the current moment and covers the current moment;

[0008] Determine the predicted line loss value at the preset time according to the predicted value of the high frequency component at the preset time and the predicted value of the low frequency component at the preset time;

[0009] According to the predicted line loss value at the preset moment and the actual line loss value at the preset moment, it is determined whether the actual line loss value at the preset moment is abnormal.

[0010] In one embodiment, establishing a line loss value database includes:

[0011] Get the line loss value at each historical moment;

[0012] If the line loss value at any historical moment cannot be obtained, the estimated line loss value at any historical moment is obtained based on the regression interpolation method, and the estimated line loss value is used as the line loss value at any historical moment;

[0013] A line loss value database is constructed based on the line loss value at each historical moment.

[0014] In one embodiment, based on the regression interpolation method, the line loss value at any historical moment is obtained, including:

[0015] Obtain the line loss values ​​of the first preset number of historical moments before any historical moment and the line loss values ​​of the second preset number of historical moments after any historical moment;

[0016] Multiply the line loss values ​​of the first preset number of historical moments by their corresponding weights, multiply the line loss values ​​of the second preset number of historical moments by their corresponding weights, sum all the products, random factors and regression equation constants to obtain a first summation result, and use the first summation result as the line loss value at any historical moment.

[0017] In one embodiment, wavelet transform processing is performed on the line loss value at each historical moment in the line loss value database to obtain the high-frequency component and low-frequency component corresponding to the line loss value at each historical moment, including:

[0018] For any historical moment, based on each value of the integral converted into a discrete sum variable and any historical moment, a first filtering result of the first wavelet decomposition filter is obtained, the first filtering result corresponding to each value of the integral converted into a discrete sum variable is multiplied by the wavelet coefficient of the high-frequency part of the line loss value at any historical moment at the first preset number of layers, and the products corresponding to each value of the integral converted into a discrete sum variable are summed to obtain the high-frequency component corresponding to the line loss value at any historical moment;

[0019] Based on each value of the integral converted into a discrete sum of variables and any historical moment, the second filtering result of the second wavelet decomposition filter is obtained, the second filtering result corresponding to each value of the integral converted into a discrete sum of variables is multiplied by the wavelet coefficient of the low-frequency part of the line loss value at any historical moment at the second preset number of layers, the products corresponding to each value of the integral converted into a discrete sum of variables are summed, and the low-frequency component corresponding to the line loss value at any historical moment is obtained.

[0020] In one embodiment, according to the high frequency component corresponding to the line loss value at each historical moment, the predicted value of the high frequency component at a preset moment in the current time period is determined, and according to the low frequency component corresponding to the line loss value at each historical moment, the predicted value of the low frequency component at the preset moment is determined, including:

[0021] The high-frequency component corresponding to the line loss value at each historical moment is input into the autoregressive moving average model, and the predicted value of the high-frequency component at the preset moment is output;

[0022] The low-frequency component corresponding to the line loss value at each historical moment is input into the long short-term memory network model, and the predicted value of the low-frequency component at the preset moment is output.

[0023] In one embodiment, determining the predicted line loss value at the preset moment according to the predicted value of the high frequency component at the preset moment and the predicted value of the low frequency component at the preset moment includes:

[0024] Multiplying the third filtering result corresponding to each value of the integral converted into a discrete sum variable by the wavelet coefficient of the high-frequency part of the high-frequency component prediction value at the preset moment at the third preset number of layers, and summing all the products to obtain a second summation result;

[0025] Multiplying the fourth filtering result corresponding to each value of the integral converted into a discrete sum variable by the wavelet coefficient of the low-frequency part of the low-frequency component prediction value at the preset time at the fourth preset number of layers, and summing all the products to obtain a third summation result;

[0026] The predicted line loss value at the preset time is calculated according to the second summation result and the third summation result.

[0027] In a second aspect, the present application also provides a line loss abnormality detection device. The device comprises:

[0028] Establishing a module for establishing a line loss value database; the line loss value database contains the line loss value at each historical moment in a preset historical time period;

[0029] The wavelet transform module is used to perform wavelet transform processing on the line loss value at each historical moment in the line loss value database to obtain the high-frequency component and low-frequency component corresponding to the line loss value at each historical moment;

[0030] A first determination module is used to determine the high-frequency component prediction value at a preset moment in the current time period according to the high-frequency component corresponding to the line loss value at each historical moment, and to determine the low-frequency component prediction value at the preset moment according to the low-frequency component corresponding to the line loss value at each historical moment, wherein the current time period is determined by the current moment and covers the current moment;

[0031] A second determination module, used to determine the predicted line loss value at the preset moment according to the predicted value of the high frequency component at the preset moment and the predicted value of the low frequency component at the preset moment;

[0032] The third determination module is used to determine whether the actual line loss value at the preset moment is abnormal according to the predicted line loss value at the preset moment and the actual line loss value at the preset moment.

[0033] In a third aspect, the present application further provides a computer device. The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:

[0034] Establish a line loss value database; the line loss value database contains the line loss value of each historical moment in a preset historical time period; perform wavelet transform processing on the line loss value of each historical moment in the line loss value database to obtain the high-frequency component and low-frequency component corresponding to the line loss value of each historical moment; determine the high-frequency component prediction value of the preset moment in the current time period according to the high-frequency component corresponding to the line loss value of each historical moment, and determine the low-frequency component prediction value of the preset moment according to the low-frequency component corresponding to the line loss value of each historical moment, the current time period is determined by the current moment and covers the current moment; determine the predicted line loss value of the preset moment according to the predicted value of the high-frequency component at the preset moment and the predicted value of the low-frequency component at the preset moment; determine whether the actual line loss value at the preset moment is abnormal according to the predicted line loss value at the preset moment and the actual line loss value at the preset moment.

[0035] In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0036] Establish a line loss value database; the line loss value database contains the line loss value of each historical moment in a preset historical time period; perform wavelet transform processing on the line loss value of each historical moment in the line loss value database to obtain the high-frequency component and low-frequency component corresponding to the line loss value of each historical moment; determine the high-frequency component prediction value of the preset moment in the current time period according to the high-frequency component corresponding to the line loss value of each historical moment, and determine the low-frequency component prediction value of the preset moment according to the low-frequency component corresponding to the line loss value of each historical moment, the current time period is determined by the current moment and covers the current moment; determine the predicted line loss value of the preset moment according to the predicted value of the high-frequency component at the preset moment and the predicted value of the low-frequency component at the preset moment; determine whether the actual line loss value at the preset moment is abnormal according to the predicted line loss value at the preset moment and the actual line loss value at the preset moment.

[0037] In a fifth aspect, the present application further provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0038] Establish a line loss value database; the line loss value database contains the line loss value of each historical moment in a preset historical time period; perform wavelet transform processing on the line loss value of each historical moment in the line loss value database to obtain the high-frequency component and low-frequency component corresponding to the line loss value of each historical moment; determine the high-frequency component prediction value of the preset moment in the current time period according to the high-frequency component corresponding to the line loss value of each historical moment, and determine the low-frequency component prediction value of the preset moment according to the low-frequency component corresponding to the line loss value of each historical moment, the current time period is determined by the current moment and covers the current moment; determine the predicted line loss value of the preset moment according to the predicted value of the high-frequency component at the preset moment and the predicted value of the low-frequency component at the preset moment; determine whether the actual line loss value at the preset moment is abnormal according to the predicted line loss value at the preset moment and the actual line loss value at the preset moment.

[0039] The above-mentioned line loss abnormality detection method, device, computer equipment, storage medium and computer program product do not manually determine whether the line loss is abnormal, but establish a line loss value database; the line loss value database contains the line loss value at each historical moment in the preset historical time period; the line loss value at each historical moment in the line loss value database is subjected to wavelet transformation processing to obtain the high-frequency component and low-frequency component corresponding to the line loss value at each historical moment; according to the high-frequency component corresponding to the line loss value at each historical moment, the high-frequency component prediction value at the preset moment in the current time period is determined; according to the low-frequency component corresponding to the line loss value at each historical moment, the low-frequency component prediction value at the preset moment is determined, and the current time period is determined by the current moment and covers the current moment; according to the high-frequency component prediction value at the preset moment and the low-frequency component prediction value at the preset moment, the predicted line loss value at the preset moment is determined; according to the predicted line loss value at the preset moment and the real line loss value at the preset moment, whether the real line loss value at the preset moment is abnormal is determined. In this way, the labor cost is reduced, the efficiency of line loss abnormality analysis is improved, and the accuracy of line loss abnormality analysis results is improved, thereby providing a basis for reducing the line loss rate and improving the intelligent comprehensive management level of power enterprises. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 A schematic diagram of a flow chart of a line loss abnormality detection method in one embodiment;

[0041] Figure 2 A schematic diagram of a flow chart of a line loss abnormality detection method in another embodiment;

[0042] Figure 3 is a structural block diagram of a line loss abnormality detection device in one embodiment;

[0043] Figure 4 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0045] Line loss is an important indicator for evaluating the operating performance of the power transmission network. In the related technology, the line loss anomaly analysis method of the power transmission and transformation equipment is mainly: after manually screening out the high-loss and negative-loss equipment, the power supply side and the supply side are compared and analyzed, and the line loss of the equipment connected to the previous voltage level or the same voltage level is combined to determine the anomaly. The above-mentioned line loss anomaly analysis method has the following problems: First, the manual analysis process is more affected by the outside world and there are many uncontrollable factors, resulting in low accuracy of the line loss anomaly analysis results; second, a large amount of manpower costs are required; third, the efficiency is low. Therefore, there is an urgent need for a line loss anomaly detection method to solve the above problems.

[0046] It is understood that the terms "first", "second", etc. used in this application can be used in this article to describe various professional terms, but unless otherwise specified, these professional terms are not limited by these terms. These terms are only used to distinguish one professional term from another professional term. For example, without departing from the scope of this application, the first filtering result and the second filtering result can be the same or different.

[0047] In response to the problems existing in the above-mentioned related technologies, an embodiment of the present application provides a line loss anomaly detection method, which can be applied to a server, a terminal, or a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. Among them, the server can be implemented as an independent server or a server cluster composed of multiple servers. The terminal can be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, Internet of Things devices and portable wearable devices. The Internet of Things devices can be smart speakers, smart TVs, smart air conditioners, smart car-mounted devices, etc. Portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. It should be noted that the numbers of "multiple" and the like mentioned in the embodiments of the present application all refer to the number of "at least two", for example, "multiple" refers to "at least two".

[0048] In one embodiment, Figure 1 As shown, a line loss anomaly detection method is provided. This embodiment uses the method applied to a server as an example for illustration. The method includes the following steps:

[0049] 102. Establish a line loss value database; the line loss value database contains the line loss value at each historical moment in a preset historical time period.

[0050] Among them, the historical time period from the start of the operation of the power grid to the current moment can be used as the target historical time period, and any sub-time period within the target historical time period can be used as the preset historical time period. Generally, when the power grid is transmitting power normally, the line loss value of the power grid is relatively stable within a time period and basically does not change. Based on this, it can be understood that if the duration between the end moment of the preset historical time period and the start moment of the current time period satisfies the preset condition that it is less than the first preset threshold, the predicted line loss value at the preset moment will be more stable and accurate, so that the result of determining whether the actual line loss value at the preset moment is abnormal will also be more accurate. It can also be understood that the preset historical time period and the current time period may also exist at the same moment.

[0051] It should be noted that, in actual applications, the line loss value at each historical moment can be obtained through smart meter testing, and then the smart meter uploads the line loss value at each historical moment to the server.

[0052] 104. Perform wavelet transform processing on the line loss value at each historical moment in the line loss value database to obtain high-frequency components and low-frequency components corresponding to the line loss value at each historical moment.

[0053] It should be noted that wavelet transform is based on certain specific functions and expands the data signal into a series of series. It is a local transformation of time, space and frequency. Therefore, wavelet transform can perform multi-scale joint analysis on data signals in both time domain and frequency domain, so as to effectively extract information from the signal. The significance of wavelet transform lies in the ability to decompose signals at different scales, and different scales can be selected according to the actual target situation. It can decompose data information into detail information and approximate information. Among them, detail represents the high scale of the signal, that is, high-frequency information, and approximate represents the high scale of the signal, that is, low-frequency information, and the low-frequency information often contains the characteristics of the signal. Therefore, the original signal can generate two signals through two mutual filters.

[0054] Wavelet transform continuously decomposes the approximate signal, but in practical applications, the number of decomposition levels is selected according to appropriate criteria. The idea of ​​wavelet transform is the same as that of Fourier transform, except that the infinite-length trigonometric function basis is replaced by a finite-length wavelet basis that will decay. The formula of wavelet transform can be shown as follows (1):

[0055]

[0056] Among them, WT (a, τ) represents the wavelet transform function, t represents time, the scale factor is represented by a, and the translation is represented by τ. The scale factor controls the expansion and contraction of the wavelet function. Different expansion factors generate different frequency components and affect the frequency components. The translation controls the translation of the wavelet function and affects the time component, so that the wavelet can realize traversal analysis along the time axis of the signal.

[0057] 106. According to the high-frequency component corresponding to the line loss value at each historical moment, the predicted value of the high-frequency component at the preset moment in the current time period is determined. According to the low-frequency component corresponding to the line loss value at each historical moment, the predicted value of the low-frequency component at the preset moment is determined. The current time period is determined by the current moment and covers the current moment.

[0058] The current time period may include only the current time, or may include multiple preset time periods, including the current time period. This embodiment of the present application does not specifically limit this.

[0059] 108. Determine a predicted line loss value at a preset moment according to a predicted value of a high frequency component at a preset moment and a predicted value of a low frequency component at a preset moment.

[0060] Specifically, the predicted value of the high-frequency component at the preset moment and the predicted value of the low-frequency component at the preset moment may be reconstructed by wavelet to determine the predicted line loss value at the preset moment.

[0061] 110. Determine whether the actual line loss value at the preset moment is abnormal based on the predicted line loss value at the preset moment and the actual line loss value at the preset moment.

[0062] Specifically, the difference between the predicted line loss value at the preset moment and the actual line loss value at the preset moment is obtained; it is determined whether the difference is greater than a second preset threshold value, if greater, it is determined that the actual line loss value at the preset moment is abnormal, if less, it is determined that the actual line loss value at the preset moment is not abnormal. In practical applications, the actual line loss value at the preset moment can be obtained by testing a smart meter, and then the smart meter uploads the actual line loss value at the preset moment to a server.

[0063] In the above-mentioned line loss abnormality detection method, since it is not judged manually whether the line loss is abnormal, but by establishing a line loss value database; the line loss value database contains the line loss value of each historical moment in the preset historical time period; the line loss value of each historical moment in the line loss value database is subjected to wavelet transform processing to obtain the high-frequency component and low-frequency component corresponding to the line loss value of each historical moment; according to the high-frequency component corresponding to the line loss value of each historical moment, the high-frequency component prediction value of the preset moment in the current time period is determined; according to the low-frequency component corresponding to the line loss value of each historical moment, the low-frequency component prediction value of the preset moment is determined, and the current time period is determined by the current moment and covers the current moment; according to the high-frequency component prediction value of the preset moment and the low-frequency component prediction value of the preset moment, the predicted line loss value of the preset moment is determined; according to the predicted line loss value of the preset moment and the real line loss value of the preset moment, it is determined whether the real line loss value of the preset moment is abnormal. In this way, the labor cost is reduced, the efficiency of line loss abnormality analysis is improved, and the accuracy of the line loss abnormality analysis results is improved, thereby providing a basis for reducing the line loss rate and improving the intelligent comprehensive management level of power enterprises.

[0064] In one embodiment, a line loss value database is established, including: obtaining the line loss value at each historical moment; if there is a line loss value at any historical moment that cannot be obtained, then based on the regression interpolation method, obtaining the estimated line loss value at any historical moment, and using the estimated line loss value as the line loss value at any historical moment; according to the line loss value at each historical moment, a line loss value database is constructed.

[0065] It should be noted that the line loss value at any historical moment can also be determined based on other methods. Specifically, an average value can be calculated based on the line loss value of each historical moment that has been obtained, and the average value can be used as the line loss value at any historical moment. It is also possible to obtain the median or mode based on the line loss value of each historical moment that has been obtained, and use the median or mode as the line loss value at any historical moment. The line loss value at any historical moment is also determined based on linear interpolation, exponential interpolation, polynomial interpolation, and logarithmic difference, etc., and the embodiments of the present application do not specifically limit this. It should be noted that the reason why the regression interpolation method is used to determine the line loss value at any historical moment is, on the one hand, because the regression interpolation method infers an optimal regression equation based on the line loss value of each historical moment that has been obtained, and then the line loss value at any historical moment is obtained through the regression equation. The obtained line loss value at any historical moment is closer to the real data than that obtained by other methods; on the other hand, it is because the regression interpolation method is simple and efficient.

[0066] In this embodiment, since the line loss value at any historical moment is missing, the accuracy of the predicted line loss value at the preset moment will be reduced. Therefore, the line loss value at each historical moment is obtained; if there is a line loss value at any historical moment that cannot be obtained, the estimated line loss value at any historical moment is obtained based on the regression interpolation method, and the estimated line loss value is used as the line loss value at any historical moment; according to the line loss value at each historical moment, a line loss value database is constructed. In this way, the accuracy of the predicted line loss value at the preset moment is improved.

[0067] In one embodiment, based on the regression interpolation method, the line loss value at any historical moment is obtained, including: obtaining the line loss values ​​of the first preset number of historical moments before any historical moment and the line loss values ​​of the second preset number of historical moments after any historical moment; multiplying the line loss values ​​of the first preset number of historical moments by their respective corresponding weights, multiplying the line loss values ​​of the second preset number of historical moments by their respective corresponding weights, summing all the products, random factors and regression equation constants to obtain a first summation result, and using the first summation result as the line loss value at any historical moment.

[0068] Specifically, the process of obtaining the line loss value at any historical moment can refer to the following formula (2):

[0069]

[0070] It should be noted that in formula (2), y k represents the line loss value at any historical moment, α0 represents the regression equation constant, n represents the sum of the first preset number and the second preset number, α i represents the weight corresponding to the line loss value of any historical moment in the first preset number of historical moments or the second preset number of historical moments, X i represents the line loss value at any historical moment among the first preset number of historical moments or the second preset number of historical moments, and ξ represents a random factor.

[0071] In this embodiment, the line loss values ​​of the first preset number of historical moments before any historical moment and the line loss values ​​of the second preset number of historical moments after any historical moment are obtained; the line loss values ​​of the first preset number of historical moments before any historical moment are multiplied by their corresponding weights, and the line loss values ​​of the second preset number of historical moments after any historical moment are multiplied by their corresponding weights, and all products, random factors and regression equation constants are summed to obtain a first summation result, and the first summation result is used as the line loss value of any historical moment. Since the regression interpolation method is to infer an optimal regression equation based on the line loss value of each historical moment that has been obtained, and then the line loss value of any historical moment is obtained through the regression equation, the obtained line loss value of any historical moment is closer to the real data.

[0072] In one embodiment, the line loss value at each historical moment in the line loss value database is subjected to wavelet transformation processing to obtain the high-frequency component and the low-frequency component corresponding to the line loss value at each historical moment, including:

[0073] For any historical moment, based on each value of the integral converted into a discrete sum of variables and any historical moment, the first filtering result of the first wavelet decomposition filter is obtained, the first filtering result corresponding to each value of the integral converted into a discrete sum of variables is multiplied by the wavelet coefficient of the high-frequency part of the line loss value at any historical moment at the first preset number of layers, the products corresponding to each value of the integral converted into a discrete sum of variables are summed, and the high-frequency component corresponding to the line loss value at any historical moment is obtained.

[0074] The first wavelet decomposition filter may be a high-pass filter.

[0075] Specifically, the process of determining the high-frequency component corresponding to the line loss value at any historical moment can refer to the following formula (3):

[0076]

[0077] It should be noted that in formula (3), D j [f(t)] represents the high-frequency component corresponding to the line loss value at any historical moment, G(2t-m) represents the first filtering result corresponding to any value of the integral into discrete sum variables, the value range of m is 0 to 2t-1, D j-1 [f(t)] represents the wavelet coefficient of the high-frequency part of the line loss value at any historical moment at the first preset number of layers, and j-1 represents the first preset number of layers.

[0078] Based on each value of the integral converted into a discrete sum of variables and any historical moment, the second filtering result of the second wavelet decomposition filter is obtained, the second filtering result corresponding to each value of the integral converted into a discrete sum of variables is multiplied by the wavelet coefficient of the low-frequency part of the line loss value at any historical moment at the second preset number of layers, the products corresponding to each value of the integral converted into a discrete sum of variables are summed, and the low-frequency component corresponding to the line loss value at any historical moment is obtained.

[0079] The second wavelet decomposition filter may be a low-pass filter.

[0080] Specifically, the process of determining the low-frequency component corresponding to the line loss value at any historical moment can refer to the following formula (4):

[0081]

[0082] It should be noted that in formula (4), A j[f(t)] represents the low-frequency component corresponding to the line loss value at any historical moment, H(2t-m) represents the second filtering result corresponding to any value of the integral into discrete sum variables, the value range of m is 0 to 2t-1, A j-1 [f(t)] represents the wavelet coefficient of the low-frequency part of the line loss value at any historical moment at the second preset number of layers, and j-1 represents the second preset number of layers.

[0083] In this embodiment, the line loss value is gradually refined at multiple scales through the scaling and translation operation of wavelet transformation, so as to separate the high and low frequency components of the line loss value without losing the information component of the line loss value, thereby making the predicted line loss value more accurate.

[0084] In one embodiment, according to the high frequency component corresponding to the line loss value at each historical moment, the predicted value of the high frequency component at a preset moment in the current time period is determined, and according to the low frequency component corresponding to the line loss value at each historical moment, the predicted value of the low frequency component at the preset moment is determined, including:

[0085] The high-frequency component corresponding to the line loss value at each historical moment is input into the autoregressive moving average model, and the predicted value of the high-frequency component at the preset moment is output; the low-frequency component corresponding to the line loss value at each historical moment is input into the long short-term memory network model, and the predicted value of the low-frequency component at the preset moment is output.

[0086] Specifically, the low-frequency component corresponding to the line loss value at each historical moment can also be input into the autoregressive moving average model to output the low-frequency component prediction value at the preset moment. The high-frequency component corresponding to the line loss value at each historical moment can also be input into the long short-term memory network model to output the high-frequency component prediction value at the preset moment. The reason why the high-frequency component prediction value at the preset moment is obtained through the autoregressive moving average model and the low-frequency component prediction value at the preset moment is obtained through the long short-term memory network model is to improve the accuracy of the high-frequency component prediction value and the low-frequency component prediction value at the preset moment.

[0087] It should be noted that the autoregressive moving average model, referred to as the ARIMA model, is a statistical model that can be used to predict time series. The principle of the autoregressive moving average model is to transform a non-stationary time series into a stationary time series and then regress the dependent variable only on its lagged value and the present value and lagged value of the random error term to establish a model. Among the three parameters, p represents the number of lags of the time series data itself used in the prediction model, also called the Auto-Regressive term, d represents the number of differences (order) made by the time series data d to make it a stationary series, also called the Integrated term, and q represents the number of lags of the prediction error used in the prediction model, also called the MA / Moving Average term.

[0088] In the autoregressive moving average model, AR is the autoregressive part of the model, and MA is its moving average part. AR describes the relationship between the current value and the historical value, and uses the historical time data of the variable itself to predict itself. The use of the autoregressive model must meet the requirements of stationarity and must have autocorrelation, that is, the autocorrelation coefficient must be greater than or equal to 0.5. The formula for the p-order autoregressive process can be shown as follows (5):

[0089]

[0090] Among them, y t represents the current data value, p represents the order, u represents the constant term, ξ t Represents the error term. The size of p is determined by the partial correlation function, which removes the interference of the intermediate k-1 variables and measures the influence of y(tk) on y(t).

[0091] The moving average model focuses on the accumulation of error terms in the autoregressive model. The moving average method can effectively eliminate random fluctuations in the prediction. The formula of the q-order autoregressive process can be shown as follows (6):

[0092]

[0093] The size of q is determined by the autocorrelation function. The autocorrelation coefficient reflects the correlation between the values ​​of the same sequence at different time series. Its calculation formula is shown in the following formula (7):

[0094]

[0095] Among them, Cov represents covariance and Var represents variance.

[0096] The autoregressive moving average model can be expressed as follows:

[0097]

[0098] in, represents the autoregressive coefficient, θ q represents the autoregressive coefficient.

[0099] The LSTM network model is a special neural network model that can learn long-term dependencies. The LSTM network model adds a memory unit to maintain the long-term memory of the network. In the LSTM network model, these operations are implemented through a structure called gates, which is also the clever part of the LSTM network model. These gate structures make the weights of the self-loops in the LSTM network model change, controlling how much information can pass through the previous unit and what information the current unit has that can be added to the cell state and passed to the next unit. Among them, there are three types of gate structures: input gate, forget gate, and output gate. The function of the forget gate is to decide what information we will discard from the cell state. It reads the output of the previous cell and the input of the current cell, and finally outputs a value between 0 and 1 for each number in the cell state. 1 means "completely retain" and 0 means "completely discard". The processing process of the forget gate can be shown in the following formula (9):

[0100] f t =σ(W f *[h t-1 ,x t ]+b f ); (9)

[0101] Among them, σ(·) represents the sigmoid function, h t-1 It represents the output of the previous cell, x t Represents the input of the current cell, W f represents the weight corresponding to f, b f Represents the bias parameter corresponding to f.

[0102] The function of the input gate is to decide how much new information to add to the cell state. First, a sigmoid layer called the "input gate layer" decides which information needs to be updated. Then, a tanh layer generates a vector, which is used as a backup update content. Finally, these two parts are combined to update the cell state. The processing process of the input gate can be shown as the following formula (10):

[0103] i t =σ(Wi *[h t-1 ,x t ]+b i )

[0104] C' t =tanh(W C *[h t-1 ,x t ]+b C ); (10)

[0105] Among them, i represents the state value of the input gate, W i represents the weight corresponding to i, b i represents the bias parameter corresponding to i. Then, we compare the old state with f t Multiply, discard the information we are sure we need to discard, and add i t C' t We get new candidate values, which then change based on how much we decide to update each state.

[0106] The output gate determines what value to output based on the cell state. First, a sigmoid layer is run to determine which part of the cell state will be output. Next, the cell state is processed through tanh and multiplied with the output of the sigmoid gate. In the end, only the part that determines the output will be output. The processing of the output gate can be shown as the following formula (11):

[0107] o t =σ(W o *[h t-1 ,x t ]+b o )

[0108] h t =o t *tanh(C t ); (11)

[0109] Among them, h t represents the output of the LSTM unit, t represents the current time, and W o represents the corresponding weight, b i Indicates the corresponding bias parameter.

[0110] In this embodiment, the high-frequency component corresponding to the line loss value at each historical moment is input into the autoregressive moving average model to output the high-frequency component prediction value at the preset moment; the low-frequency component corresponding to the line loss value at each historical moment is input into the long short-term memory network model to output the low-frequency component prediction value at the preset moment. This improves the accuracy of the high-frequency component prediction value and the low-frequency component prediction value at the preset moment.

[0111] In one embodiment, the predicted line loss value at the preset moment is determined based on the predicted value of the high-frequency component at the preset moment and the predicted value of the low-frequency component at the preset moment, including: multiplying the third filtering result corresponding to each value of the integral converted into a discrete sum variable with the wavelet coefficient of the high-frequency part of the predicted value of the high-frequency component at the preset moment at the third preset number of layers, summing all the products to obtain a second summation result; multiplying the fourth filtering result corresponding to each value of the integral converted into a discrete sum variable with the wavelet coefficient of the low-frequency part of the predicted value of the low-frequency component at the preset moment at the fourth preset number of layers, summing all the products to obtain a third summation result; and calculating the predicted line loss value at the preset moment based on the second summation result and the third summation result.

[0112] Specifically, the calculation steps of the predicted line loss value at the preset time can refer to the following formula (12):

[0113]

[0114] It should be noted that in formula (12), C j [f(t)] represents the predicted line loss value at the preset time, h(t-2k) represents the fourth filtering result corresponding to any value of the integral into discrete sum variables, A j+1 [f(t)] represents the wavelet coefficient of the low-frequency part of the low-frequency component prediction value at the preset time under the fourth preset number of layers, g(2t-m) represents the third filtering result corresponding to any value of the integral into discrete sum variables, D j+1 [f(t)] represents the wavelet coefficient of the high-frequency part of the high-frequency component prediction value at the preset time at the third preset number of layers, and the value range of m is 0 to 2t-1.

[0115] In one embodiment, the specific process of determining whether the actual line loss value at a preset time is abnormal can be as follows: Figure 2 As shown, the present application embodiment does not make any specific limitation to this. Figure 2 As shown, a line loss value database can also be constructed based on the line loss value at each historical moment and the actual line loss value at a preset moment. It can be understood that Figure 2 The preprocessing process in is also the process of "obtaining the estimated value of line loss at any historical moment based on regression interpolation method". It should be noted that Figure 2 For the explanations of the terms and steps involved, please refer to the explanations of the terms and steps in the above embodiments, which will not be repeated here.

[0116] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0117] Based on the same inventive concept, the embodiment of the present application also provides a line loss anomaly detection device for implementing the line loss anomaly detection method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more line loss anomaly detection device embodiments provided below can refer to the limitations of the line loss anomaly detection method above, and will not be repeated here.

[0118] In one embodiment, Figure 3 As shown, a line loss abnormality detection device is provided, including: an establishment module 302, a wavelet transformation module 304, a first determination module 306, a second determination module 308 and a third determination module 310, wherein:

[0119] Establishing module 302, for establishing a line loss value database; the line loss value database contains the line loss value at each historical moment in a preset historical time period;

[0120] The wavelet transform module 304 is used to perform wavelet transform processing on the line loss value at each historical moment in the line loss value database to obtain the high-frequency component and the low-frequency component corresponding to the line loss value at each historical moment;

[0121] A first determination module 306 is used to determine a high-frequency component prediction value at a preset moment in a current time period according to a high-frequency component corresponding to a line loss value at each historical moment, and to determine a low-frequency component prediction value at a preset moment according to a low-frequency component corresponding to a line loss value at each historical moment, wherein the current time period is determined by and covers the current moment;

[0122] A second determination module 308, configured to determine a predicted line loss value at a preset moment according to a predicted value of a high frequency component at a preset moment and a predicted value of a low frequency component at a preset moment;

[0123] The third determination module 310 is used to determine whether the actual line loss value at the preset moment is abnormal according to the predicted line loss value at the preset moment and the actual line loss value at the preset moment.

[0124] In one embodiment, the establishment module 302 includes:

[0125] A first acquisition unit, used to acquire the line loss value at each historical moment;

[0126] A second acquisition unit is used to acquire an estimated value of the line loss value at any historical moment based on a regression interpolation method if the line loss value at any historical moment cannot be acquired, and use the estimated value of the line loss value as the line loss value at any historical moment;

[0127] The construction unit is used to construct a line loss value database according to the line loss value at each historical moment.

[0128] In one embodiment, the second acquisition unit includes:

[0129] An acquisition subunit, used to acquire line loss values ​​of a first preset number of historical moments before any historical moment and line loss values ​​of a second preset number of historical moments after any historical moment;

[0130] A determination subunit is provided, which is used to multiply the line loss values ​​of a first preset number of historical moments by their respective corresponding weights, and to multiply the line loss values ​​of a second preset number of historical moments by their respective corresponding weights, and to sum all the products, random factors and regression equation constants to obtain a first summation result, and to use the first summation result as the line loss value of any historical moment.

[0131] In one embodiment, the wavelet transform module 304 includes:

[0132] A third acquisition unit is used to acquire, for any historical moment, a first filtering result of a first wavelet decomposition filter based on each value of the integral converted into a discrete sum variable and any historical moment, multiply the first filtering result corresponding to each value of the integral converted into a discrete sum variable by the wavelet coefficient of the high-frequency part of the line loss value at any historical moment at a first preset number of layers, sum the products corresponding to each value of the integral converted into a discrete sum variable, and obtain a high-frequency component corresponding to the line loss value at any historical moment;

[0133] The fourth acquisition unit is used to obtain the second filtering result of the second wavelet decomposition filter based on each value of the integral converted into a discrete sum variable and any historical moment, multiply the second filtering result corresponding to each value of the integral converted into a discrete sum variable with the wavelet coefficient of the low-frequency part of the line loss value at any historical moment at a second preset number of layers, sum the products corresponding to each value of the integral converted into a discrete sum variable, and obtain the low-frequency component corresponding to the line loss value at any historical moment.

[0134] In one embodiment, the first determining module 306 includes:

[0135] A first output unit, used to input the high-frequency component corresponding to the line loss value at each historical moment into the autoregressive moving average model, and output the predicted value of the high-frequency component at a preset moment;

[0136] The second output unit is used to input the low-frequency component corresponding to the line loss value at each historical moment into the long short-term memory network model, and output the predicted value of the low-frequency component at a preset moment.

[0137] In one embodiment, the second determining module 308 includes:

[0138] A first summing unit is used to multiply the third filtering result corresponding to each value of the integral-converted discrete sum variable by the wavelet coefficient of the high-frequency part of the high-frequency component prediction value at a preset time at a third preset number of layers, and sum all the products to obtain a second summing result;

[0139] A second summing unit is used to multiply the fourth filtering result corresponding to each value of the integral-converted discrete sum variable by the wavelet coefficient of the low-frequency part of the low-frequency component prediction value at a preset time at a fourth preset number of layers, and sum all the products to obtain a third summing result;

[0140] The calculation unit is used to calculate the predicted line loss value at a preset time according to the second summation result and the third summation result.

[0141] Each module in the above-mentioned line loss anomaly detection device can be implemented in whole or in part by software, hardware, or a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0142] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 4 As shown. The computer device includes a processor, a memory and a network interface connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the actual line loss value at a preset time, the predicted line loss value at a preset time and the data in the line loss value database. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a line loss anomaly detection method is implemented.

[0143] Those skilled in the art will understand that Figure 4The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0144] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0145] Establishing a line loss value database; the line loss value database contains the line loss value at each historical moment in a preset historical time period;

[0146] Perform wavelet transform processing on the line loss value at each historical moment in the line loss value database to obtain the high-frequency component and low-frequency component corresponding to the line loss value at each historical moment;

[0147] According to the high-frequency component corresponding to the line loss value at each historical moment, the predicted value of the high-frequency component at the preset moment in the current time period is determined; according to the low-frequency component corresponding to the line loss value at each historical moment, the predicted value of the low-frequency component at the preset moment is determined, and the current time period is determined by the current moment and covers the current moment;

[0148] Determine the predicted line loss value at the preset time according to the predicted value of the high frequency component at the preset time and the predicted value of the low frequency component at the preset time;

[0149] According to the predicted line loss value at the preset moment and the actual line loss value at the preset moment, it is determined whether the actual line loss value at the preset moment is abnormal.

[0150] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0151] Get the line loss value at each historical moment;

[0152] If the line loss value at any historical moment cannot be obtained, the estimated line loss value at any historical moment is obtained based on the regression interpolation method, and the estimated line loss value is used as the line loss value at any historical moment;

[0153] A line loss value database is constructed based on the line loss value at each historical moment.

[0154] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0155] Obtain the line loss values ​​of the first preset number of historical moments before any historical moment and the line loss values ​​of the second preset number of historical moments after any historical moment;

[0156] Multiply the line loss values ​​of the first preset number of historical moments by their corresponding weights, multiply the line loss values ​​of the second preset number of historical moments by their corresponding weights, sum all the products, random factors and regression equation constants to obtain a first summation result, and use the first summation result as the line loss value at any historical moment.

[0157] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0158] For any historical moment, based on each value of the integral converted into a discrete sum variable and any historical moment, a first filtering result of the first wavelet decomposition filter is obtained, the first filtering result corresponding to each value of the integral converted into a discrete sum variable is multiplied by the wavelet coefficient of the high-frequency part of the line loss value at any historical moment at the first preset number of layers, and the products corresponding to each value of the integral converted into a discrete sum variable are summed to obtain the high-frequency component corresponding to the line loss value at any historical moment;

[0159] Based on each value of the integral converted into a discrete sum of variables and any historical moment, the second filtering result of the second wavelet decomposition filter is obtained, the second filtering result corresponding to each value of the integral converted into a discrete sum of variables is multiplied by the wavelet coefficient of the low-frequency part of the line loss value at any historical moment at the second preset number of layers, the products corresponding to each value of the integral converted into a discrete sum of variables are summed, and the low-frequency component corresponding to the line loss value at any historical moment is obtained.

[0160] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0161] The high-frequency component corresponding to the line loss value at each historical moment is input into the autoregressive moving average model, and the predicted value of the high-frequency component at the preset moment is output;

[0162] The low-frequency component corresponding to the line loss value at each historical moment is input into the long short-term memory network model, and the predicted value of the low-frequency component at the preset moment is output.

[0163] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0164] Multiplying the third filtering result corresponding to each value of the integral converted into a discrete sum variable by the wavelet coefficient of the high-frequency part of the high-frequency component prediction value at the preset moment at the third preset number of layers, and summing all the products to obtain a second summation result;

[0165] Multiplying the fourth filtering result corresponding to each value of the integral converted into a discrete sum variable by the wavelet coefficient of the low-frequency part of the low-frequency component prediction value at the preset time at the fourth preset number of layers, and summing all the products to obtain a third summation result;

[0166] The predicted line loss value at the preset time is calculated according to the second summation result and the third summation result.

[0167] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0168] Establishing a line loss value database; the line loss value database contains the line loss value at each historical moment in a preset historical time period;

[0169] Perform wavelet transform processing on the line loss value at each historical moment in the line loss value database to obtain the high-frequency component and low-frequency component corresponding to the line loss value at each historical moment;

[0170] According to the high-frequency component corresponding to the line loss value at each historical moment, the predicted value of the high-frequency component at the preset moment in the current time period is determined; according to the low-frequency component corresponding to the line loss value at each historical moment, the predicted value of the low-frequency component at the preset moment is determined, and the current time period is determined by the current moment and covers the current moment;

[0171] Determine the predicted line loss value at the preset time according to the predicted value of the high frequency component at the preset time and the predicted value of the low frequency component at the preset time;

[0172] According to the predicted line loss value at the preset moment and the actual line loss value at the preset moment, it is determined whether the actual line loss value at the preset moment is abnormal.

[0173] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0174] Get the line loss value at each historical moment;

[0175] If the line loss value at any historical moment cannot be obtained, the estimated line loss value at any historical moment is obtained based on the regression interpolation method, and the estimated line loss value is used as the line loss value at any historical moment;

[0176] A line loss value database is constructed based on the line loss value at each historical moment.

[0177] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0178] Obtain the line loss values ​​of the first preset number of historical moments before any historical moment and the line loss values ​​of the second preset number of historical moments after any historical moment;

[0179] Multiply the line loss values ​​of the first preset number of historical moments by their corresponding weights, multiply the line loss values ​​of the second preset number of historical moments by their corresponding weights, sum all the products, random factors and regression equation constants to obtain a first summation result, and use the first summation result as the line loss value at any historical moment.

[0180] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0181] For any historical moment, based on each value of the integral converted into a discrete sum variable and any historical moment, a first filtering result of the first wavelet decomposition filter is obtained, the first filtering result corresponding to each value of the integral converted into a discrete sum variable is multiplied by the wavelet coefficient of the high-frequency part of the line loss value at any historical moment at the first preset number of layers, and the products corresponding to each value of the integral converted into a discrete sum variable are summed to obtain the high-frequency component corresponding to the line loss value at any historical moment;

[0182] Based on each value of the integral converted into a discrete sum of variables and any historical moment, the second filtering result of the second wavelet decomposition filter is obtained, the second filtering result corresponding to each value of the integral converted into a discrete sum of variables is multiplied by the wavelet coefficient of the low-frequency part of the line loss value at any historical moment at the second preset number of layers, the products corresponding to each value of the integral converted into a discrete sum of variables are summed, and the low-frequency component corresponding to the line loss value at any historical moment is obtained.

[0183] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0184] The high-frequency component corresponding to the line loss value at each historical moment is input into the autoregressive moving average model, and the predicted value of the high-frequency component at the preset moment is output;

[0185] The low-frequency component corresponding to the line loss value at each historical moment is input into the long short-term memory network model, and the predicted value of the low-frequency component at the preset moment is output.

[0186] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0187] Multiplying the third filtering result corresponding to each value of the integral converted into a discrete sum variable by the wavelet coefficient of the high-frequency part of the high-frequency component prediction value at the preset moment at the third preset number of layers, and summing all the products to obtain a second summation result;

[0188] Multiplying the fourth filtering result corresponding to each value of the integral converted into a discrete sum variable by the wavelet coefficient of the low-frequency part of the low-frequency component prediction value at the preset time at the fourth preset number of layers, and summing all the products to obtain a third summation result;

[0189] The predicted line loss value at the preset time is calculated according to the second summation result and the third summation result.

[0190] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0191] Establishing a line loss value database; the line loss value database contains the line loss value at each historical moment in a preset historical time period;

[0192] Perform wavelet transform processing on the line loss value at each historical moment in the line loss value database to obtain the high-frequency component and low-frequency component corresponding to the line loss value at each historical moment;

[0193] According to the high-frequency component corresponding to the line loss value at each historical moment, the predicted value of the high-frequency component at the preset moment in the current time period is determined; according to the low-frequency component corresponding to the line loss value at each historical moment, the predicted value of the low-frequency component at the preset moment is determined, and the current time period is determined by the current moment and covers the current moment;

[0194] Determine the predicted line loss value at the preset time according to the predicted value of the high frequency component at the preset time and the predicted value of the low frequency component at the preset time;

[0195] According to the predicted line loss value at the preset moment and the actual line loss value at the preset moment, it is determined whether the actual line loss value at the preset moment is abnormal.

[0196] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0197] Get the line loss value at each historical moment;

[0198] If the line loss value at any historical moment cannot be obtained, the estimated line loss value at any historical moment is obtained based on the regression interpolation method, and the estimated line loss value is used as the line loss value at any historical moment;

[0199] A line loss value database is constructed based on the line loss value at each historical moment.

[0200] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0201] Obtain the line loss values ​​of the first preset number of historical moments before any historical moment and the line loss values ​​of the second preset number of historical moments after any historical moment;

[0202] Multiply the line loss values ​​of the first preset number of historical moments by their corresponding weights, multiply the line loss values ​​of the second preset number of historical moments by their corresponding weights, sum all the products, random factors and regression equation constants to obtain a first summation result, and use the first summation result as the line loss value at any historical moment.

[0203] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0204] For any historical moment, based on each value of the integral converted into a discrete sum variable and any historical moment, a first filtering result of the first wavelet decomposition filter is obtained, the first filtering result corresponding to each value of the integral converted into a discrete sum variable is multiplied by the wavelet coefficient of the high-frequency part of the line loss value at any historical moment at the first preset number of layers, and the products corresponding to each value of the integral converted into a discrete sum variable are summed to obtain the high-frequency component corresponding to the line loss value at any historical moment;

[0205] Based on each value of the integral converted into a discrete sum of variables and any historical moment, the second filtering result of the second wavelet decomposition filter is obtained, the second filtering result corresponding to each value of the integral converted into a discrete sum of variables is multiplied by the wavelet coefficient of the low-frequency part of the line loss value at any historical moment at the second preset number of layers, the products corresponding to each value of the integral converted into a discrete sum of variables are summed, and the low-frequency component corresponding to the line loss value at any historical moment is obtained.

[0206] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0207] The high-frequency component corresponding to the line loss value at each historical moment is input into the autoregressive moving average model, and the predicted value of the high-frequency component at the preset moment is output;

[0208] The low-frequency component corresponding to the line loss value at each historical moment is input into the long short-term memory network model, and the predicted value of the low-frequency component at the preset moment is output.

[0209] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0210] Multiplying the third filtering result corresponding to each value of the integral converted into a discrete sum variable by the wavelet coefficient of the high-frequency part of the high-frequency component prediction value at the preset moment at the third preset number of layers, and summing all the products to obtain a second summation result;

[0211] Multiplying the fourth filtering result corresponding to each value of the integral converted into a discrete sum variable by the wavelet coefficient of the low-frequency part of the low-frequency component prediction value at the preset time at the fourth preset number of layers, and summing all the products to obtain a third summation result;

[0212] The predicted line loss value at the preset time is calculated according to the second summation result and the third summation result.

[0213] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0214] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0215] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A line loss abnormality detection method, characterized in that: The method comprises: Establishing a line loss value database; the line loss value database contains the line loss value at each historical moment in a preset historical time period; For any historical moment, based on each value of the integral converted into a discrete sum variable and the any historical moment, obtain a first filtering result of a high-pass filter, multiply the first filtering result corresponding to each value of the integral converted into a discrete sum variable by the wavelet coefficient of the high-frequency part of the line loss value at the any historical moment at the first preset number of layers, sum the products corresponding to each value of the integral converted into a discrete sum variable, and obtain the high-frequency component corresponding to the line loss value at the any historical moment; based on each value of the integral converted into a discrete sum variable and the any historical moment, obtain a second filtering result of a low-pass filter, multiply the second filtering result corresponding to each value of the integral converted into a discrete sum variable by the wavelet coefficient of the low-frequency part of the line loss value at the any historical moment at the second preset number of layers, sum the products corresponding to each value of the integral converted into a discrete sum variable, and obtain the low-frequency component corresponding to the line loss value at the any historical moment; According to the high-frequency component corresponding to the line loss value at each historical moment, a predicted value of the high-frequency component at a preset moment in the current time period is determined; according to the low-frequency component corresponding to the line loss value at each historical moment, a predicted value of the low-frequency component at the preset moment is determined, and the current time period is determined by the current moment and covers the current moment; The third filtering result corresponding to each value of the integral converted into a discrete sum variable is multiplied by the wavelet coefficient of the high frequency part of the high frequency component prediction value at the preset moment at the third preset number of layers, and all the products are summed to obtain a second summation result; the fourth filtering result corresponding to each value of the integral converted into a discrete sum variable is multiplied by the wavelet coefficient of the low frequency part of the low frequency component prediction value at the preset moment at the fourth preset number of layers, and all the products are summed to obtain a third summation result; the predicted line loss value at the preset moment is calculated according to the second summation result and the third summation result; According to the predicted line loss value at the preset moment and the actual line loss value at the preset moment, it is determined whether the actual line loss value at the preset moment is abnormal.

2. The method according to claim 1, characterized in that The establishing of the line loss value database comprises: Get the line loss value at each historical moment; If the line loss value at any historical moment cannot be obtained, then based on the regression interpolation method, an estimated line loss value at any historical moment is obtained, and the estimated line loss value is used as the line loss value at any historical moment; The line loss value database is constructed according to the line loss value at each historical moment.

3. The method according to claim 2, characterized in that The obtaining of the line loss value at any historical moment based on the regression interpolation method includes: Obtaining line loss values ​​of a first preset number of historical moments before any historical moment and line loss values ​​of a second preset number of historical moments after any historical moment; Multiply the line loss values ​​of the first preset number of historical moments by their corresponding weights, multiply the line loss values ​​of the second preset number of historical moments by their corresponding weights, sum all the products, random factors and regression equation constants to obtain a first summation result, and use the first summation result as the line loss value of any historical moment.

4. The method according to claim 1, characterized in that: The step of determining the high-frequency component prediction value at a preset moment in the current time period according to the high-frequency component corresponding to the line loss value at each historical moment, and determining the low-frequency component prediction value at the preset moment according to the low-frequency component corresponding to the line loss value at each historical moment, includes: Inputting the high-frequency component corresponding to the line loss value at each historical moment into the autoregressive moving average model, and outputting the predicted value of the high-frequency component at the preset moment; The low-frequency component corresponding to the line loss value at each historical moment is input into the long short-term memory network model, and the predicted value of the low-frequency component at the preset moment is output.

5. A line loss abnormality detection device, characterized in that: The device comprises: An establishment module is used to establish a line loss value database; the line loss value database contains the line loss value at each historical moment in a preset historical time period; A wavelet transform module is used to obtain, for any historical moment, a high-pass filtering result of a high-pass filter based on each value of the integral converted into a discrete sum variable and the any historical moment, multiply the high-pass filtering result corresponding to each value of the integral converted into a discrete sum variable with the wavelet coefficient of the high-frequency part of the line loss value at any historical moment at the first preset number of layers, sum the products corresponding to each value of the integral converted into a discrete sum variable, and obtain the high-frequency component corresponding to the line loss value at any historical moment; based on each value of the integral converted into a discrete sum variable and the any historical moment, obtain a low-pass filtering result of a low-pass filter based on each value of the integral converted into a discrete sum variable and the any historical moment, multiply the low-pass filtering result corresponding to each value of the integral converted into a discrete sum variable with the wavelet coefficient of the low-frequency part of the line loss value at any historical moment at the second preset number of layers, sum the products corresponding to each value of the integral converted into a discrete sum variable, and obtain the low-frequency component corresponding to the line loss value at any historical moment; A first determination module is used to determine a high-frequency component prediction value at a preset moment in a current time period according to a high-frequency component corresponding to a line loss value at each historical moment, and to determine a low-frequency component prediction value at the preset moment according to a low-frequency component corresponding to a line loss value at each historical moment, wherein the current time period is determined by and covers the current moment; The second determination module is used to multiply the third filtering result corresponding to each value of the integral converted into a discrete sum variable with the wavelet coefficient of the high frequency part of the high frequency component prediction value at the preset moment at the third preset number of layers, and sum all the products to obtain a second summation result; multiply the fourth filtering result corresponding to each value of the integral converted into a discrete sum variable with the wavelet coefficient of the low frequency part of the low frequency component prediction value at the preset moment at the fourth preset number of layers, and sum all the products to obtain a third summation result; calculate the predicted line loss value at the preset moment according to the second summation result and the third summation result, The third determination module is used to determine whether the actual line loss value at the preset moment is abnormal according to the predicted line loss value at the preset moment and the actual line loss value at the preset moment.

6. The device according to claim 5, characterized in that The establishment module comprises: A first acquisition unit, used to acquire the line loss value at each historical moment; A second acquisition unit is used to acquire an estimated value of the line loss value at any historical moment based on a regression interpolation method if the line loss value at any historical moment cannot be acquired, and use the estimated value of the line loss value as the line loss value at any historical moment; The construction unit is used to construct a line loss value database according to the line loss value at each historical moment.

7. The device according to claim 6, characterized in that The second acquisition unit includes: An acquisition subunit, used to acquire line loss values ​​of a first preset number of historical moments before any historical moment and line loss values ​​of a second preset number of historical moments after any historical moment; A determination subunit is provided, which is used to multiply the line loss values ​​of a first preset number of historical moments by their respective corresponding weights, and to multiply the line loss values ​​of a second preset number of historical moments by their respective corresponding weights, and to sum all the products, random factors and regression equation constants to obtain a first summation result, and to use the first summation result as the line loss value of any historical moment.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images