Method, device and computer program product for determining line loss rate of distribution network

Abstract

The present application provides a method, device, and computer program product for determining the line loss rate of a distribution network. The method comprises: obtaining multiple important line loss indicator data, where the important line loss indicator data is indicator data that affects the line loss rate of the distribution network; inputting the multiple important line loss indicator data into a line loss rate determination model, using the line loss rate determination model to perform empirical mode decomposition on the important line loss indicator data to obtain multiple time series modal components, and superimposing and reconstructing the line loss rate components corresponding to each time series modal component to output a target line loss rate for the distribution network; and executing a loss reduction operation when the target line loss rate is greater than a line loss rate threshold, where the loss reduction operation is an operation to reduce the line loss rate of the distribution network. This method solves the problem of inaccurate line loss rate prediction in the prior art for distribution network.

Description

Technical Field

[0001] The present invention relates to the field of distribution network line loss rate prediction, and in particular to a method for determining a distribution network line loss rate, a determination device, a computer-readable storage medium, and a computer program product. Background Art

[0002] Line loss rate is a key factor affecting the economic benefits of power companies. A large number of distributed generation (DG) are connected to distribution networks. However, due to factors such as weather, their unique timing characteristics change the way distribution networks operate. Power generation at different times can affect line losses. Therefore, considering the timing characteristics of DG can improve the accuracy of line loss rate predictions for distribution networks, which is of great significance for improving the economic benefits of power companies.

[0003] Currently, line loss rate prediction methods for distribution networks with DGs are primarily categorized into two categories: mathematical modeling and machine learning. Mathematical modeling is the most traditional method for predicting line loss rates. It typically relies on power system theory and empirical formulas, analyzing and modeling factors such as the grid's physical structure, load characteristics, and transmission lines to predict line loss rates. Machine learning methods utilize a data-driven approach, building neural network models and training algorithms to identify correlations between line loss rates and corresponding indicator data, thereby predicting line loss rates. Mathematical modeling, based on power system theory and empirical formulas, offers a certain degree of transparency and interpretability. However, DG systems may exhibit significant temporal variations due to factors such as weather conditions and load fluctuations, and the models often fail to capture these variations, limiting their accuracy. Machine learning methods can handle complex data and their nonlinear relationships without requiring mathematical models. Common BP neural network models have a simple structure but cannot predict line loss based on DG temporal characteristics. LSTM neural network models are more complex and typically require significant computational resources and time. Furthermore, they may not effectively capture sequential characteristics when the dataset is too small or too short. Summary of the Invention

[0004] The main purpose of this application is to provide a method for determining the line loss rate of a distribution network, a determination device, a computer-readable storage medium and a computer program product, so as to at least solve the problem of inaccurate prediction of the line loss rate of a distribution network in the prior art.

[0005] To achieve the above-mentioned purpose, according to one aspect of the present application, a method for determining the line loss rate of a distribution network is provided, comprising: obtaining a plurality of important line loss indicator data, wherein the important line loss indicator data are indicator data that affect the line loss rate of the distribution network; inputting the plurality of important line loss indicator data into a line loss rate determination model, using the line loss rate determination model to perform empirical mode decomposition on the important line loss indicator data to obtain a plurality of time series modal components, and superimposing and reconstructing the line loss rate components corresponding to each of the time series modal components, and outputting a target line loss rate of the distribution network, wherein the line loss rate determination model is a convergence model obtained by performing modal decomposition on a first sample training set and inputting it into a neural network for training, wherein the first sample training set includes a plurality of groups of first training data, each group of the first training data includes sample important indicator data and a first sample line loss rate corresponding to the sample important indicator data; when the target line loss rate is greater than a line loss rate threshold, performing a loss reduction operation, wherein the loss reduction operation is an operation of reducing the line loss rate of the distribution network.

[0006] Optionally, obtaining multiple important line loss indicator data includes: obtaining all indicator data of the distribution network, the indicator data being data corresponding to various indicator types, the indicator types including at least feeder power supply radius, daily load rate fluctuation coefficient, maximum temperature, minimum temperature and weather type; performing noise reduction processing on all the indicator data to obtain first indicator data corresponding to each indicator type; performing normalization processing on all the first indicator data to obtain second indicator data corresponding to each indicator type; inputting the second indicator data into a linear regression model in sequence to obtain feature weights corresponding to each second indicator data, the linear regression model being a convergence model obtained by inputting a second sample training set into a linear regression equation for training, the second sample training set including multiple groups of second training data, each group of the second training data including sample second indicator data, and a second sample line loss rate and sample feature weight corresponding to the sample second indicator data; sorting all the feature weights from large to small to obtain a weight sequence; selecting the second indicator data corresponding to the feature weights whose sequence numbers in the weight sequence are less than or equal to a set number, and determining them as the important line loss indicator data.

[0007] Optionally, performing noise reduction processing on all the indicator data to obtain first indicator data corresponding to each indicator type includes: removing null values ​​in all the indicator data to obtain first noise reduction indicator data corresponding to each indicator type; calculating the deviation degree of each indicator type corresponding to the first noise reduction indicator data according to a first formula, wherein the first formula is The determination of the distribution network line loss rate represents the deviation degree, i represents the first

[0008] A number of noise reduction index data, X irepresents the i-th first noise reduction index data corresponding to the current index type, μ represents the mean of all the first noise reduction index data corresponding to the current index type, and σ represents the standard deviation of the first noise reduction index data corresponding to the current index type; when the absolute value of the deviation degree is greater than or equal to the deviation threshold, the first noise reduction index data is eliminated; when the absolute value of the deviation degree is less than the deviation threshold, the first noise reduction index data is determined as the first index data.

[0009] Optionally, before the second indicator data is sequentially input into the linear regression model, the method further includes: constructing an initial linear regression model, which is the linear regression model to be trained; an analysis step, inputting the sample second indicator data in multiple groups of current second training data into the initial linear regression model for linear regression analysis to obtain the corresponding current first sample line loss rate, and the current second training data is the second training data of any group in the second sample training set; an optimization step, using the mean square error as the first loss function, and using the least squares method to optimize the first loss function, the first loss function is used to measure the difference between the current first sample line loss rate and the second sample line loss rate of the sample second indicator data; repeating the analysis step and the optimization step in sequence until the first loss value of the first loss function reaches the minimum, obtaining the linear regression model, and determining the regression coefficient of the linear regression model as the sample feature weight, and the first loss value is the difference between the current first sample line loss rate and the second sample line loss rate of the sample second indicator data.

[0010] Optionally, before inputting the plurality of said important line loss indicator data into the line loss rate determination model, the method further comprises: constructing an initial line loss rate determination model, the initial line loss rate determination model comprising a modal decomposition module and a neural network module, the modal decomposition module being used to perform modal decomposition on the input data, and the neural network module being used to calculate the modal decomposed data to obtain output data; a first input step, inputting the said sample important indicator data in a plurality of groups of current first training data into the initial line loss rate determination model, the said current first training data being any group of the first training data in the first sample training set; a decomposition step, performing empirical mode decomposition on the said sample important indicator data to obtain a plurality of sample modal components; a second input step, inputting the said plurality of groups of current first training data into the initial line loss rate determination model, The sample modal components are input from the input layer of the neural network, and are calculated through multiple hidden layers of the neural network to obtain sample line loss rate components corresponding to each sample modal component; a reconstruction step is performed to superimpose and reconstruct the sample line loss rate components corresponding to each sample modal component to obtain the current second sample line loss rate; the first input step, the decomposition step, the second input step and the reconstruction step are repeated at least once in sequence until the number of iterations reaches the maximum number of iterations or the second loss value of the second loss function is less than the second loss threshold, to obtain the line loss rate determination model, the second loss function is the mean square error function between the first sample line loss rate and the current second sample line loss rate, and the second loss value is the value of the mean square error function.

[0011] Optionally, multiple sample modal components are input from the input layer of the neural network, and calculated through multiple hidden layers of the neural network to obtain sample line loss rate components corresponding to each sample modal component, including: skip-connecting the hidden layers, passing the first output information of the first hidden layer to the second hidden layer for calculation, and obtaining the second output information of the second hidden layer, the first hidden layer is the hidden layer currently being calculated, the second hidden layer is the hidden layer not adjacent to the first hidden layer, the first output information is the output of the first hidden layer and the input of the second hidden layer; the first output information and the second output information are added to obtain the sample line loss rate components corresponding to each sample modal component.

[0012] Optionally, after outputting the target line loss rate of the distribution network, the method further includes: when the target line loss rate is less than or equal to the line loss rate threshold, continuously monitoring the operation of the distribution network, and updating the line loss rate determination model periodically or in real time.

[0013] According to another aspect of the present application, a device for determining a line loss rate of a distribution network is provided, the device comprising: an acquisition unit for acquiring a plurality of important line loss indicator data, wherein the important line loss indicator data are indicator data affecting the line loss rate of the distribution network; an input / output unit for inputting the plurality of important line loss indicator data into a line loss rate determination model, so as to use the line loss rate determination model to perform empirical mode decomposition on the important line loss indicator data to obtain a plurality of time series modal components, and to superimpose and reconstruct line loss rate components corresponding to each of the time series modal components to output a target line loss rate of the distribution network, wherein the line loss rate determination model is a convergence model obtained by performing modal decomposition on a first sample training set and then inputting the resultant data into a neural network for training, the first sample training set comprising a plurality of groups of first training data, each group of the first training data comprising sample important indicator data and a first sample line loss rate corresponding to the sample important indicator data; and an execution unit for executing a loss reduction operation when the target line loss rate is greater than a line loss rate threshold, the loss reduction operation being an operation of reducing the line loss rate of the distribution network.

[0014] According to another aspect of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute any one of the methods described.

[0015] According to another aspect of the present application, a computer program product is provided, comprising computer instructions, wherein when the computer instructions are executed by a processor, any one of the methods described above is implemented.

[0016] Applying the technical solution of the present application, in a method for determining the line loss rate of a distribution network, first, a plurality of important line loss indicator data are obtained, and the above-mentioned important line loss indicator data are indicator data that affect the line loss rate of the distribution network; then, the plurality of the above-mentioned important line loss indicator data are input into a line loss rate determination model, so as to use the above-mentioned line loss rate determination model to perform empirical mode decomposition on the above-mentioned important line loss indicator data to obtain a plurality of time series modal components, and the line loss rate components corresponding to each of the above-mentioned time series modal components are superimposed and reconstructed, and the target line loss rate of the above-mentioned distribution network is output. The above-mentioned line loss rate determination model is a convergence model obtained by performing modal decomposition on a first sample training set and inputting it into a neural network for training. The above-mentioned first sample training set includes multiple groups of first training data, and each group of the above-mentioned first training data includes sample important indicator data and a first sample line loss rate corresponding to the above-mentioned sample important indicator data; finally, when the above-mentioned target line loss rate is greater than the line loss rate threshold, a loss reduction operation is performed, and the above-mentioned loss reduction operation is an operation of reducing the above-mentioned line loss rate of the above-mentioned distribution network. This application selects multiple important line loss indicator data that have a significant impact on the line loss rate and uses them as input data for the line loss rate determination model. It then uses the empirical mode decomposition method to decompose each important line loss indicator data into multiple modal components. A neural network is then used to predict the line loss rate for each modal component of the important line loss indicator data, obtaining the line loss rate component for each indicator data. All line loss rate components are then superimposed and reconstructed to obtain the target line loss rate. This allows for distribution network line loss rate prediction that takes into account the timing characteristics of DGs, enabling more flexible adaptation to changes in the DG system and making the prediction results more accurate and reliable. This application addresses the issue of inaccurate distribution network line loss rate prediction in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A hardware structure block diagram of a mobile terminal for executing a method for determining a line loss rate of a distribution network provided in an embodiment of the present application is shown;

[0018] Figure 2 A schematic flow chart of a method for determining a line loss rate of a distribution network provided in accordance with an embodiment of the present application is shown;

[0019] Figure 3 A schematic flow chart of a method for determining a line loss rate of a distribution network according to an embodiment of the present application is shown;

[0020] Figure 4 A flow chart of an empirical mode decomposition method provided according to an embodiment of the present application is shown;

[0021] Figure 5 A structural diagram of a line loss rate determination model for a distribution network provided in accordance with an embodiment of the present application is shown;

[0022] Figure 6A graph showing an actual daily line loss rate and a predicted value provided according to an embodiment of the present application is shown;

[0023] Figure 7 A structural block diagram of a device for determining a line loss rate of a distribution network provided in accordance with an embodiment of the present application is shown.

[0024] The above drawings include the following reference numerals:

[0025] 102. Processor; 104. Memory; 106. Transmission device; 108. Input / output device. DETAILED DESCRIPTION

[0026] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0027] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention 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 in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of this application.

[0028] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described here. In addition, the terms "including" and "having" and any of their variations 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 clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0029] As introduced in the background technology, the BP neural network model commonly used in the prior art has a simple structure, but cannot predict line loss based on the DG timing characteristics; the LSTM network neural network model has a relatively complex structure, usually requires a large amount of computing resources and time, and may also be unable to effectively capture sequence characteristics because the data set is too small or too short. In order to solve the problem of inaccurate prediction of distribution network line loss rate, the embodiments of the present application provide a method for determining the distribution network line loss rate, a determination device, a computer-readable storage medium and a computer program product.

[0030] The technical solutions in the embodiments of the present invention will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present invention.

[0031] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Taking running on a mobile terminal as an example, Figure 1 FIG. 1 is a hardware structure block diagram of a mobile terminal for determining a distribution network line loss rate according to an embodiment of the present invention. Figure 1 As shown, the mobile terminal may include one or more ( Figure 1 Only one is shown) a processor 102 (the processor 102 may include but is not limited to a microprocessor MCU or a programmable logic device FPGA and other processing devices) and a memory 104 for storing data, wherein the mobile terminal may also include a transmission device 106 and an input and output device 108 for communication functions. It will be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the mobile terminal. Figure 1 More or fewer components than shown, or with Figure 1 Different configurations shown.

[0032] The memory 104 can be used to store computer programs, such as application software programs and modules, such as the computer program corresponding to the method for determining the power distribution network line loss rate in the embodiment of the present invention. The processor 102 executes the computer program stored in the memory 104 to execute various functional applications and data processing, thereby implementing the above-mentioned method. The memory 104 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include memory remotely located from the processor 102, and these remote memories can be connected to the mobile terminal via a network. Examples of such networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof. The transmission device 106 is used to receive or transmit data via a network. Specific examples of such networks may include a wireless network provided by the mobile terminal's telecommunications provider. In one example, the transmission device 106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to enable communication with the Internet. In one example, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0033] In this embodiment, a method for determining the line loss rate of a distribution network running on a mobile terminal, a computer terminal or a similar computing device 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 can be executed in an order different from that shown here.

[0034] Figure 2 FIG. 1 is a flow chart of a method for determining a distribution network line loss rate according to an embodiment of the present application. Figure 2 As shown, the method includes the following steps:

[0035] Step S201 : Acquire a plurality of important line loss indicator data, where the important line loss indicator data is indicator data that affects the line loss rate of the distribution network.

[0036] Specifically, multiple indicators that have an impact on the line loss rate are selected. After data preprocessing, the daily line loss rate and its related indicator data are obtained, and then multiple important indicators that have the greatest impact on the line loss rate are further selected, namely the above-mentioned important line loss indicator data.

[0037] In step S202, the plurality of important line loss indicator data are input into a line loss rate determination model, and the line loss rate determination model is used to perform empirical mode decomposition on the important line loss indicator data to obtain a plurality of time series modal components, and the line loss rate components corresponding to each of the time series modal components are superimposed and reconstructed to output the target line loss rate of the distribution network. The line loss rate determination model is a convergence model obtained by performing modal decomposition on the first sample training set and inputting it into a neural network for training. The first sample training set includes a plurality of groups of first training data, and each group of the first training data includes sample important indicator data and the first sample line loss rate corresponding to the sample important indicator data.

[0038] Specifically, multiple key line loss indicator data sets are input into a line loss rate determination model to determine the line loss rate. Within the line loss rate determination model, empirical mode decomposition (EMD) is performed on each key line loss indicator data set. This EMD of the line loss rate components effectively separates different time series modal components, enabling better analysis of the influencing factors and changing patterns of the line loss rate. This process yields n time series modal components corresponding to each key line loss indicator data set. Each time series modal component is then predicted using a neural network to obtain the corresponding line loss rate component. The line loss rate components corresponding to all these time series modal components are then superimposed and reconstructed to obtain the target line loss rate, which is then output.

[0039] Step S203: When the target line loss rate is greater than the line loss rate threshold, a loss reduction operation is performed, where the loss reduction operation is an operation of reducing the line loss rate of the distribution network.

[0040] Specifically, when the target line loss rate is greater than the line loss rate threshold, it indicates that the line loss rate is too high, and quick measures need to be taken to repair and optimize the rate in order to reduce the line loss rate.

[0041] In this embodiment, first, a plurality of important line loss indicator data are obtained, and the above-mentioned important line loss indicator data are indicator data that affect the line loss rate of the distribution network; then, the plurality of the above-mentioned important line loss indicator data are input into the line loss rate determination model, and the above-mentioned line loss rate determination model is used to perform empirical mode decomposition on the above-mentioned important line loss indicator data to obtain a plurality of time series modal components, and the line loss rate components corresponding to each of the above-mentioned time series modal components are superimposed and reconstructed, and the target line loss rate of the above-mentioned distribution network is output. The above-mentioned line loss rate determination model is a convergence model obtained by performing modal decomposition on the first sample training set and inputting it into the neural network for training. The above-mentioned first sample training set includes multiple groups of first training data, and each group of the above-mentioned first training data includes sample important indicator data and the first sample line loss rate corresponding to the above-mentioned sample important indicator data; finally, when the above-mentioned target line loss rate is greater than the line loss rate threshold, a loss reduction operation is performed, and the above-mentioned loss reduction operation is an operation of reducing the above-mentioned line loss rate of the above-mentioned distribution network. This application selects multiple important line loss indicator data that have a significant impact on the line loss rate and uses them as input data for the line loss rate determination model. It then uses the empirical mode decomposition method to decompose each important line loss indicator data into multiple modal components. A neural network is then used to predict the line loss rate for each modal component of the important line loss indicator data, obtaining the line loss rate component for each indicator data. All line loss rate components are then superimposed and reconstructed to obtain the target line loss rate. This allows for distribution network line loss rate prediction that takes into account the timing characteristics of DGs, enabling more flexible adaptation to changes in the DG system and making the prediction results more accurate and reliable. This application addresses the issue of inaccurate distribution network line loss rate prediction in the prior art.

[0042] In order to enable those skilled in the art to more clearly understand the technical solution of the present application, the implementation process of the method for determining the line loss rate of the distribution network of the present application will be described in detail below with reference to specific embodiments.

[0043] In order to improve data quality and thus improve the accuracy of the line loss rate determination model, in an optional implementation manner, the above step S201 includes:

[0044] Step S2011: Acquire all indicator data of the distribution network, where the indicator data corresponds to various indicator types, including at least feeder power supply radius, daily load factor fluctuation coefficient, maximum temperature, minimum temperature, and weather type.

[0045] Step S2012: performing noise reduction processing on all the above indicator data to obtain first indicator data corresponding to each of the above indicator types;

[0046] Step S2013, normalizing all the first indicator data to obtain second indicator data corresponding to each of the indicator types;

[0047] Step S2014: sequentially input the second indicator data into a linear regression model to obtain a feature weight corresponding to each of the second indicator data. The linear regression model is a convergent model obtained by inputting the second sample training set into a linear regression equation for training. The second sample training set includes multiple groups of second training data, each group of the second training data includes sample second indicator data, and a second sample line loss rate and a sample feature weight corresponding to the sample second indicator data.

[0048] Step S2015: sort all the above feature weights from largest to smallest to obtain a weight sequence;

[0049] Step S2016: Select the second indicator data corresponding to the feature weights whose sequence numbers in the weight sequence are less than or equal to a set number, and determine them as the important line loss indicator data.

[0050] In the above embodiment, if Figure 3 As shown, first obtain the daily line loss rate and historical indicator data, such as the feeder power supply radius, total length, number of dedicated transformers, dedicated transformer capacity, cabling rate, proportion of energy-saving distribution transformers, integrated power, area power, dedicated transformer power, distributed online active power, distributed online reactive power, proportion of distributed online power, number of distributed power sources, voltage qualification rate, proportion of public transformers with low power factor, proportion of public transformers with three-phase imbalance, average load rate of distribution lines, proportion of electricity consumption of large industrial power consumption (only dedicated transformers are counted), peak reverse active power of local power plants, average reverse active power of local power plants, valley reverse active power of local power plants, peak reverse active power of local power plants, total reverse active power of line gateway meters, peak reverse active power of line gateway meters, daily load rate fluctuation coefficient, maximum temperature, minimum temperature, and weather type as indicator types affecting the line loss rate, and then obtain the indicator data of these indicator types for preprocessing. Due to measurement, collection, and transmission failures, null values ​​or large deviations in the data may occur, which will have an adverse effect on model training. Therefore, noise removal is required to obtain the above-mentioned first indicator data. Different types of data indicators usually have different dimensions, which may affect the model training results. In order to eliminate such effects, data normalization is required. The min-max normalization method is used to map the original indicator data to a value in the range [0, 1]. The min-max normalization conversion function is defined as: Where: are the maximum and minimum values ​​of the first indicator data corresponding to the indicator type, X i is the initial value of the data, is the standard value of the data. After normalization, the second indicator data is obtained. Next, indicator selection is performed. The coefficient (weight) of each feature in linear regression analysis can reflect the degree of influence of the feature on the target variable. A larger absolute coefficient generally indicates a greater impact on the prediction result. Therefore, this analysis can assess the impact of each indicator on the line loss rate. The top 10 indicator types with the greatest impact on the line loss rate (which can be set according to actual conditions) are selected, and the data corresponding to the top 10 indicator types are used as the important line loss indicator data.

[0051] In order to reduce data deviation, in an optional implementation manner, the above step S2021 includes:

[0052] Step S20211: remove all null values ​​in the above indicator data to obtain first noise reduction indicator data corresponding to each of the above indicator types;

[0053] Step S20212, calculating the deviation of each indicator type corresponding to the first noise reduction indicator data according to the first formula, the first formula is Z represents the above deviation, i represents the number of the above first noise reduction index data, X i represents the i-th first noise reduction index data corresponding to the current index type, represents the mean of all the first noise reduction index data corresponding to the current index type, and σ represents the standard deviation of the first noise reduction index data corresponding to the current index type;

[0054] Step S20213: if the absolute value of the deviation is greater than or equal to the deviation threshold, the first noise reduction index data is eliminated;

[0055] Step S20214: When the absolute value of the deviation is less than the deviation threshold, the first noise reduction index data is determined as the first index data.

[0056] In the above embodiment, the null value data is first removed to obtain the first noise reduction index data, and then the Z-score method is used to test the deviation of the first noise reduction index data: Z is the Z-score (i.e., the aforementioned degree of deviation), which indicates the degree of deviation in the first noise reduction index data. When |Z| ≥ 3, the first noise reduction index data is considered to have a significant deviation and should be eliminated. Therefore, any first noise reduction index data that does not meet the requirements is eliminated, while any remaining first noise reduction index data that meets the requirements is retained and determined as the first index data.

[0057] In order to improve the data validity during line loss rate prediction, in an optional implementation manner, before the above step S2014, the method further includes:

[0058] Step S301, constructing an initial linear regression model, where the initial linear regression model is the linear regression model to be trained;

[0059] Step S302, an analysis step, inputting the sample second indicator data in multiple sets of current second training data into the initial linear regression model for linear regression analysis to obtain the corresponding current first sample line loss rate, where the current second training data is any set of second training data in the second sample training set;

[0060] Step S303, an optimization step, using the mean square error as a first loss function and optimizing the first loss function using the least squares method, wherein the first loss function is used to measure the difference between the current first sample line loss rate and the second sample line loss rate of the sample second indicator data;

[0061] Step S304, repeat the above analysis step and the above optimization step in sequence until the first loss value of the above first loss function reaches the minimum, obtain the above linear regression model, and determine the regression coefficient of the above linear regression model as the above sample feature weight, and the above first loss value is the difference between the above current first sample line loss rate and the above second sample line loss rate of the above sample second indicator data.

[0062] In the above embodiment, the regression coefficient (weight) of each indicator type in the linear regression analysis can reflect the degree of influence of the indicator type on the target variable (i.e., line loss rate). A larger absolute coefficient usually indicates that the indicator type has a greater impact on the prediction result, so the linear regression analysis can evaluate the degree of influence of each indicator type on the line loss rate. First, a linear regression model is built to analyze the relationship between the original indicator data and the line loss rate. There are many options for linear regression models, such as simple linear regression model, multivariate linear regression model, polynomial regression model, ridge regression model, etc. As long as the linear regression analysis can be implemented, the mean square error (MSE) is used as the loss function, that is, the first loss function mentioned above, to measure the actual line loss rate y in the sample. i (i.e. the second sample line loss rate mentioned above) and the predicted line loss rate y i (i.e. the current first sample line loss rate mentioned above), where: The Ordinary Least Squares (OLS) method is then used to optimize the mean squared error (MSE) to find the regression coefficient that minimizes the first loss function. By optimizing the loss function, the difference between the current first and second sample line loss rates can be reduced, improving the model's fit and prediction accuracy. The final linear regression model is obtained until the loss function reaches a minimum, and the regression coefficient is used as the sample feature weight.

[0063] In order to improve the prediction accuracy and stability of the line loss rate, in an optional implementation manner, before the above step S202, the method further includes:

[0064] Step S401: constructing an initial line loss rate determination model, the initial line loss rate determination model including a modal decomposition module and a neural network module, the modal decomposition module being used to perform modal decomposition on input data, and the neural network module being used to calculate output data from the modal decomposition data;

[0065] Step S402, a first input step, inputting the sample important indicator data in multiple sets of current first training data into the initial line loss rate determination model, wherein the current first training data is any set of the first training data in the first sample training set;

[0066] Step S403, a decomposition step, performing empirical mode decomposition on the above sample important indicator data to obtain multiple sample modal components;

[0067] Step S404, a second input step, inputs the plurality of sample modal components from the input layer of the neural network, performs calculations through the plurality of hidden layers of the neural network, and obtains a sample line loss rate component corresponding to each of the sample modal components;

[0068] Step S405, a reconstruction step, superimposing and reconstructing the sample line loss rate components corresponding to the above sample modal components to obtain the current second sample line loss rate;

[0069] Step S406, repeat the above-mentioned first input step, the above-mentioned decomposition step, the above-mentioned second input step and the above-mentioned reconstruction step at least once in sequence until the number of iterations reaches the maximum number of iterations or the second loss value of the second loss function is less than the second loss threshold, and obtain the above-mentioned line loss rate determination model, the above-mentioned second loss function is the mean square error function between the above-mentioned first sample line loss rate and the above-mentioned current second sample line loss rate, and the above-mentioned second loss value is the numerical value of the above-mentioned mean square error function.

[0070] In the above embodiment, an initial line loss rate determination model is first constructed. The initial line loss rate determination model consists of two parts: empirical mode decomposition and neural network. The parameter settings of the neural network are shown in Table 1.

[0071] Table 1

[0072]

[0073] The neural network forward propagates the learned data features, and the sample modal components pass through the hidden layer, and finally the current second sample line loss rate is obtained from the output layer. Then, the mean square error between the current second sample line loss rate output by the output layer and the true value (i.e. the first sample line loss rate mentioned above) is used as the loss function, and the result of the loss function is used to evaluate the degree of fit of the model, and backpropagation is performed to promote the network optimization training accuracy. The gradient descent method is used to iteratively optimize the loss function to find the minimum value. In this process, parameters such as weights and bias values ​​are continuously updated, and the value of the loss function is constantly changing, with an overall decreasing trend. When the set value is reached, the learning ends. The way to adjust the weight is: w ij (t+1)=w ij (t)+ηδ p y p , where η is the gain phase, δ p is the output error (also known as the second loss value), which can be expressed as y p (i) is the current second sample line loss rate, y m (i) is the line loss rate of the first sample. By iteratively executing the input, decomposition, calculation, and reconstruction steps multiple times, the line loss rate determination model can be gradually optimized to obtain the final line loss rate determination model.

[0074] Empirical Mode Decomposition (EMD) is a novel method for processing non-stationary signals. It decomposes a signal into multiple intrinsic mode functions (IMFs) based on the data's inherent time-scale characteristics and then superimposes these functions. Without pre-setting any basis functions, it adaptively performs principal component analysis of the signal, making it suitable for analyzing both nonlinear and non-stationary signals and linear and stationary signals. For a given signal, EMD first extracts local features by finding its local maxima and minima. Using these local maxima and minima, it gradually constructs the envelope of the local maxima (upper envelope) and the local minima (lower envelope). The average of these upper and lower envelopes is calculated to obtain the signal's local mean function. This local mean function is subtracted from the original signal to obtain a one-dimensional first-order component, an IMF. A similar operation is then performed on this IMF to obtain a second IMF. This process is iterated until the remaining components become residuals or monotonic functions that approach zero. Adding these IMFs together reconstructs the original signal. Process of EMD decomposition algorithm Figure 4As shown in Figure 1, the IMF must satisfy the following two conditions: (1) the number of local extreme points and zero crossings in the function must be equal or differ by at most one over the entire time range; (2) at any point in time, the average of the upper and lower envelopes must be zero. In practice, the mean of the upper and lower envelopes cannot be zero. Usually, when the following equation is satisfied, the mean of the envelope is considered to satisfy the IMF's zero mean condition 1. Where h k (t) is the kth IMF, and ε is the screening threshold, which is generally between 0.2 and 0.3.

[0075] For example, a distribution network containing DG in a certain area is selected, and the feeder power supply radius, total length, number of special variables, distributed online reactive power, proportion of distributed online power, number of distributed power sources, average load rate of distribution lines, maximum temperature, minimum temperature, and weather type are used as line loss rate prediction indicators. The daily line loss rate and historical data of 29 indicators from November 2023 to February 2024 are collected. After data preprocessing, the final total number of samples is 13,650. 80% of the samples are randomly selected as training sets and 20% of the samples are used as test sets. The processed data set is input into the line loss rate determination model, and the number of model training times is set to 1,000. The training set is used for model training, and the test set is used to test the daily line loss rate of the distribution network containing BG. Finally, the predicted daily line loss rate obtained from the test set is compared with the original line loss rate data. Figure 6 Based on the above line loss rate determination model and the traditional BP neural network model, daily line loss rate prediction is performed on the same data set. The comparison of the mean square error (MSE) and root mean square error (RMSE) of the two models is shown in Table 2.

[0076] Table 2

[0077]

[0078] After comparison, the prediction ability of the improved line loss rate determination model has been significantly improved. It can be concluded that EMD-BP can better learn the timing characteristics of DG, flexibly adapt to the changes of DG system, and effectively predict the daily line loss rate of the distribution network containing DG.

[0079] In order to reduce the difficulty of model training, in an optional implementation, the above step S404 includes:

[0080] Step S4041: Performing a skip connection on the hidden layers, transferring the first output information of the first hidden layer to the second hidden layer for calculation, and obtaining the second output information of the second hidden layer. The first hidden layer is the hidden layer currently being calculated, and the second hidden layer is the hidden layer not adjacent to the first hidden layer. The first output information is the output of the first hidden layer and the input of the second hidden layer.

[0081] Step S4042: Add the first output information and the second output information to obtain a sample line loss rate component corresponding to each of the sample modal components.

[0082] In the above-mentioned embodiments, neural networks generally consist of an input layer, hidden layers, and an output layer. Layers are often fully connected, meaning that every neuron in each layer is connected to all neurons in the next layer, while neurons within a layer are not connected to each other. However, due to limitations in the model structure, this model not only fails to effectively learn time series but also suffers from vanishing gradients and other training difficulties during training. Therefore, to address these issues and improve the accuracy of predicting line loss rates in distribution networks containing DGs, a network optimization based on the traditional BP neural network was performed, introducing skip connections. Figure 5 This is a schematic diagram of the structure of the line loss rate determination model. In order to alleviate the gradient disappearance, the output information F(x1) of hidden layer 1 is directly transmitted to hidden layer n, and a jump connection is performed to ensure that the original information is not lost due to nonlinear transformation. Finally, the output layer information F(xn) is added to F(x1) to output the expected data. For example, for any modal component, through the network jump connection, the output vector y of hidden layer 1 is p1 , pass hidden layer 1 directly to hidden layer 3, the output vector of hidden layer 3 is y p2 , so the final output vector y of this modal component is p y p =y p1 +y p2 ,in, x i is the i-th modal component, w i represents the feature weight, and n represents the total number of modal components.

[0083] In order to avoid abnormal line loss rate of the distribution network, in an optional implementation manner, after the above step S202, the method further includes:

[0084] Step S501: When the target line loss rate is less than or equal to the line loss rate threshold, the operation of the distribution network is continuously monitored, and the line loss rate determination model is updated periodically or in real time.

[0085] In the above embodiment, when the above target line loss rate is less than or equal to the above line loss rate threshold, it means that the current line loss rate of the distribution network is normal, but it is still necessary to continuously monitor the operation of the above distribution network to prevent abnormal conditions in the distribution network, and it is also necessary to update the above line loss rate determination model regularly or in real time, continuously optimize the operation strategy and control parameters of the distribution network, improve the operation efficiency and performance of the system, reduce energy waste, and reduce operating costs.

[0086] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0087] The embodiment of the present application also provides a device for determining the line loss rate of a distribution network. It should be noted that the device for determining the line loss rate of a distribution network in the embodiment of the present application can be used to execute the method for determining the line loss rate of a distribution network provided in the embodiment of the present application. The device is used to implement the above-mentioned embodiments and preferred implementation methods, and those that have been explained will not be repeated here. As used below, the term "module" can implement a combination of software and / or hardware for a predetermined function. Although the device described in the following embodiments is preferably implemented in software, implementation in hardware, or a combination of software and hardware, is also possible and conceived.

[0088] The following introduces the device for determining the line loss rate of the distribution network provided in an embodiment of the present application.

[0089] Figure 7 FIG is a structural block diagram of a device for determining a distribution network line loss rate according to an embodiment of the present application. Figure 7 As shown, the device includes:

[0090] The acquisition unit 10 is configured to acquire a plurality of important line loss indicator data, where the important line loss indicator data are indicator data that affect the line loss rate of the distribution network.

[0091] Specifically, multiple indicators that have an impact on the line loss rate are selected. After data preprocessing, the daily line loss rate and its related indicator data are obtained, and then multiple important indicators that have the greatest impact on the line loss rate are further selected, namely the above-mentioned important line loss indicator data.

[0092] The input-output unit 20 is used to input multiple of the above-mentioned important line loss indicator data into the line loss rate determination model, so as to use the above-mentioned line loss rate determination model to perform empirical mode decomposition on the above-mentioned important line loss indicator data to obtain multiple time series modal components, and superimpose and reconstruct the line loss rate components corresponding to each of the above-mentioned time series modal components to output the target line loss rate of the above-mentioned distribution network. The above-mentioned line loss rate determination model is a convergence model obtained by performing modal decomposition on the first sample training set and inputting it into the neural network for training. The above-mentioned first sample training set includes multiple groups of first training data, and each group of the above-mentioned first training data includes sample important indicator data and the first sample line loss rate corresponding to the above-mentioned sample important indicator data.

[0093] Specifically, multiple key line loss indicator data sets are input into a line loss rate determination model to determine the line loss rate. Within the line loss rate determination model, empirical mode decomposition (EMD) is performed on each key line loss indicator data set. This EMD of the line loss rate components effectively separates different time series modal components, enabling better analysis of the influencing factors and changing patterns of the line loss rate. This process yields n time series modal components corresponding to each key line loss indicator data set. Each time series modal component is then predicted using a neural network to obtain the corresponding line loss rate component. The line loss rate components corresponding to all these time series modal components are then superimposed and reconstructed to obtain the target line loss rate, which is then output.

[0094] The execution unit 30 is configured to execute a loss reduction operation when the target line loss rate is greater than a line loss rate threshold, wherein the loss reduction operation is an operation of reducing the line loss rate of the distribution network.

[0095] Specifically, when the target line loss rate is greater than the line loss rate threshold, it indicates that the line loss rate is too high, and quick measures need to be taken to repair and optimize the rate in order to reduce the line loss rate.

[0096] In this embodiment, the acquisition unit is used to obtain multiple important line loss indicator data, and the above-mentioned important line loss indicator data are indicator data that affect the line loss rate of the distribution network; the input and output unit is used to input the multiple important line loss indicator data into the line loss rate determination model, so as to use the above-mentioned line loss rate determination model to perform empirical mode decomposition on the above-mentioned important line loss indicator data to obtain multiple time series modal components, and superimpose and reconstruct the line loss rate components corresponding to each of the above-mentioned time series modal components, and output the target line loss rate of the above-mentioned distribution network. The above-mentioned line loss rate determination model is a convergence model obtained by performing modal decomposition on the first sample training set and inputting it into the neural network for training. The above-mentioned first sample training set includes multiple groups of first training data, and each group of the above-mentioned first training data includes sample important indicator data and the first sample line loss rate corresponding to the above-mentioned sample important indicator data; the execution unit is used to perform a loss reduction operation when the above-mentioned target line loss rate is greater than the line loss rate threshold. The above-mentioned loss reduction operation is an operation of reducing the above-mentioned line loss rate of the above-mentioned distribution network. This application selects multiple important line loss indicator data that have a significant impact on the line loss rate and uses them as input data for the line loss rate determination model. It then uses the empirical mode decomposition method to decompose each important line loss indicator data into multiple modal components. A neural network is then used to predict the line loss rate for each modal component of the important line loss indicator data, obtaining the line loss rate component for each indicator data. All line loss rate components are then superimposed and reconstructed to obtain the target line loss rate. This allows for distribution network line loss rate prediction that takes into account the timing characteristics of DGs, enabling more flexible adaptation to changes in the DG system and making the prediction results more accurate and reliable. This application addresses the issue of inaccurate distribution network line loss rate prediction in the prior art.

[0097] In order to improve data quality and thus improve the accuracy of the line loss rate determination model, in an optional implementation manner, the acquisition unit includes:

[0098] An acquisition module is configured to acquire all indicator data of the distribution network, wherein the indicator data corresponds to various indicator types, and the indicator types include at least feeder power supply radius, daily load factor fluctuation coefficient, maximum temperature, minimum temperature, and weather type;

[0099] A noise reduction module performs noise reduction processing on all the above indicator data to obtain first indicator data corresponding to each of the above indicator types;

[0100] A normalization module normalizes all the first indicator data to obtain second indicator data corresponding to each of the above indicator types;

[0101] An input module sequentially inputs the second indicator data into a linear regression model to obtain a feature weight corresponding to each of the second indicator data. The linear regression model is a convergent model obtained by inputting the second sample training set into a linear regression equation for training. The second sample training set includes multiple groups of second training data, each group of the second training data includes sample second indicator data, and a second sample line loss rate and a sample feature weight corresponding to the sample second indicator data.

[0102] The sorting module sorts all the above feature weights from large to small to obtain a weight sequence;

[0103] The selection module selects the second indicator data corresponding to the feature weights whose sequence numbers in the weight sequence are less than or equal to a set number, and determines them as the important indicator data of line loss.

[0104] In the above embodiment, if Figure 3As shown, first obtain the daily line loss rate and historical indicator data, such as the feeder power supply radius, total length, number of dedicated transformers, dedicated transformer capacity, cabling rate, proportion of energy-saving distribution transformers, integrated power, area power, dedicated transformer power, distributed online active power, distributed online reactive power, proportion of distributed online power, number of distributed power sources, voltage qualification rate, proportion of public transformers with low power factor, proportion of public transformers with three-phase imbalance, average load rate of distribution lines, proportion of electricity consumption of large industrial power consumption (only dedicated transformers are counted), peak reverse active power of local power plants, average reverse active power of local power plants, valley reverse active power of local power plants, peak reverse active power of local power plants, total reverse active power of line gateway meters, peak reverse active power of line gateway meters, daily load rate fluctuation coefficient, maximum temperature, minimum temperature, and weather type as indicator types affecting the line loss rate, and then obtain the indicator data of these indicator types for preprocessing. Due to measurement, collection, and transmission failures, null values ​​or large deviations in the data may occur, which will have an adverse effect on model training. Therefore, noise removal is required to obtain the above-mentioned first indicator data. Different types of data indicators usually have different dimensions, which may affect the model training results. In order to eliminate such effects, data normalization is required. The min-max normalization method is used to map the original indicator data to a value in the range [0, 1]. The min-max normalization conversion function is defined as: Where: are the maximum and minimum values ​​of the first indicator data corresponding to the indicator type, X i is the initial value of the data, is the standard value of the data. After normalization, the second indicator data is obtained. Next, indicator selection is performed. The coefficient (weight) of each feature in linear regression analysis can reflect the degree of influence of the feature on the target variable. A larger absolute coefficient generally indicates a greater impact on the prediction result. Therefore, this analysis can assess the impact of each indicator on the line loss rate. The top 10 indicator types with the greatest impact on the line loss rate (which can be set according to actual conditions) are selected, and the data corresponding to the top 10 indicator types are used as the important line loss indicator data.

[0105] In order to reduce data deviation, in an optional embodiment, the noise reduction module includes:

[0106] A removal submodule removes all null values ​​in the above indicator data to obtain first noise reduction indicator data corresponding to each of the above indicator types;

[0107] The calculation submodule calculates the deviation of each indicator type corresponding to the first noise reduction indicator data according to the first formula. The first formula is: Z represents the above deviation, i represents the number of the above first noise reduction index data, X irepresents the i-th first noise reduction index data corresponding to the current index type, represents the mean of all the first noise reduction index data corresponding to the current index type, and σ represents the standard deviation of the first noise reduction index data corresponding to the current index type;

[0108] A removal submodule, which removes the first noise reduction index data when the absolute value of the deviation is greater than or equal to the deviation threshold;

[0109] The determination submodule determines the first noise reduction index data as the first index data when the absolute value of the deviation is less than the deviation threshold.

[0110] In the above embodiment, the null value data is first removed to obtain the first noise reduction index data, and then the Z-score method is used to test the deviation of the first noise reduction index data: Z is the Z-score (i.e., the aforementioned degree of deviation), which indicates the degree of deviation in the first noise reduction index data. When |Z| ≥ 3, the first noise reduction index data is considered to have a significant deviation and should be eliminated. Therefore, any first noise reduction index data that does not meet the requirements is eliminated, while any remaining first noise reduction index data that meets the requirements is retained and determined as the first index data.

[0111] In order to improve the data validity when predicting line loss rate, in an optional embodiment, the device further includes:

[0112] A first construction unit is configured to construct an initial linear regression model before sequentially inputting the second indicator data into the linear regression model, wherein the initial linear regression model is the linear regression model to be trained;

[0113] an analyzing unit, configured to execute an analyzing step, inputting the sample second indicator data from a plurality of sets of current second training data into the initial linear regression model for linear regression analysis, and obtaining a corresponding current first sample line loss rate, wherein the current second training data is the second training data from any set of the second sample training set;

[0114] an optimization unit, configured to perform an optimization step, use a mean square error as a first loss function, and optimize the first loss function using a least squares method, wherein the first loss function is used to measure a difference between the current first sample line loss rate and the second sample line loss rate of the sample second indicator data;

[0115] The first repeating unit is used to repeat the above-mentioned analysis step and the above-mentioned optimization step in sequence until the first loss value of the above-mentioned first loss function reaches the minimum, thereby obtaining the above-mentioned linear regression model, and determining the regression coefficient of the above-mentioned linear regression model as the above-mentioned sample feature weight. The above-mentioned first loss value is the difference between the above-mentioned current first sample line loss rate and the above-mentioned second sample line loss rate of the above-mentioned sample second indicator data.

[0116] In the above embodiment, the regression coefficient (weight) of each indicator type in the linear regression analysis can reflect the degree of influence of the indicator type on the target variable (i.e., line loss rate). A larger absolute coefficient usually indicates that the indicator type has a greater impact on the prediction result, so the linear regression analysis can evaluate the degree of influence of each indicator type on the line loss rate. First, a linear regression model is built to analyze the relationship between the original indicator data and the line loss rate. There are many options for linear regression models, such as simple linear regression model, multivariate linear regression model, polynomial regression model, ridge regression model, etc. As long as the linear regression analysis can be implemented, the mean square error (MSE) is used as the loss function, that is, the first loss function mentioned above, to measure the actual line loss rate y in the sample. i (i.e. the second sample line loss rate mentioned above) and the predicted line loss rate y i (i.e. the current first sample line loss rate mentioned above), where: The Ordinary Least Squares (OLS) method is then used to optimize the mean squared error (MSE) to find the regression coefficient that minimizes the first loss function. By optimizing the loss function, the difference between the current first and second sample line loss rates can be reduced, improving the model's fit and prediction accuracy. The final linear regression model is obtained until the loss function reaches a minimum, and the regression coefficient is used as the sample feature weight.

[0117] In order to improve the prediction accuracy and stability of the line loss rate, in an optional embodiment, the device further includes:

[0118] a second construction unit, configured to construct an initial line loss rate determination model before inputting the plurality of important line loss indicator data into the line loss rate determination model, wherein the initial line loss rate determination model includes a modal decomposition module and a neural network module, wherein the modal decomposition module is configured to perform modal decomposition on the input data, and the neural network module is configured to calculate the modal decomposition data to obtain output data;

[0119] A first input unit is configured to execute a first input step, inputting the sample important indicator data in a plurality of sets of current first training data into the initial line loss rate determination model, wherein the current first training data is any set of the first training data in the first sample training set;

[0120] A decomposition unit is used to perform the decomposition step, perform empirical mode decomposition on the important indicator data of the sample, and obtain multiple sample modal components;

[0121] A second input unit is configured to execute a second input step, inputting the plurality of sample modal components from the input layer of the neural network, performing calculations through the plurality of hidden layers of the neural network, and obtaining a sample line loss rate component corresponding to each of the sample modal components;

[0122] a reconstruction unit, configured to perform a reconstruction step, superimpose and reconstruct the sample line loss rate components corresponding to the above-mentioned sample modal components to obtain a current second sample line loss rate;

[0123] The second repetition unit is used to repeatedly execute the above-mentioned first input step, the above-mentioned decomposition step, the above-mentioned second input step and the above-mentioned reconstruction step in sequence at least once until the number of iterations reaches the maximum number of iterations or the second loss value of the second loss function is less than the second loss threshold, thereby obtaining the above-mentioned line loss rate determination model, the above-mentioned second loss function is the mean square error function between the above-mentioned first sample line loss rate and the above-mentioned current second sample line loss rate, and the above-mentioned second loss value is the numerical value of the above-mentioned mean square error function.

[0124] In the above embodiment, an initial line loss rate determination model is first constructed. The initial line loss rate determination model consists of two parts: empirical mode decomposition and neural network. The parameter settings of the neural network are shown in Table 1.

[0125] Table 1

[0126]

[0127] The neural network forward propagates the learned data features, and the sample modal components pass through the hidden layer, and finally the current second sample line loss rate is obtained from the output layer. Then, the mean square error between the current second sample line loss rate output by the output layer and the true value (i.e. the first sample line loss rate mentioned above) is used as the loss function, and the result of the loss function is used to evaluate the degree of fit of the model, and backpropagation is performed to promote the network optimization training accuracy. The gradient descent method is used to iteratively optimize the loss function to find the minimum value. In this process, parameters such as weights and bias values ​​are continuously updated, and the value of the loss function is constantly changing, with an overall decreasing trend. When the set value is reached, the learning ends. The way to adjust the weight is: w ij (t+1)=w ij (t)+ηδ p y p , where η is the gain phase, δ p is the output error (also known as the second loss value), which can be expressed as y p (i) is the current second sample line loss rate, y m(i) is the line loss rate of the first sample. By iteratively executing the input, decomposition, calculation, and reconstruction steps multiple times, the line loss rate determination model can be gradually optimized to obtain the final line loss rate determination model.

[0128] Empirical Mode Decomposition (EMD) is a novel method for processing non-stationary signals. It decomposes a signal into multiple intrinsic mode functions (IMFs) based on the data's inherent time-scale characteristics and then superimposes these functions. Without pre-setting any basis functions, it adaptively performs principal component analysis of the signal, making it suitable for analyzing both nonlinear and non-stationary signals and linear and stationary signals. For a given signal, EMD first extracts local features by finding its local maxima and minima. Using these local maxima and minima, it gradually constructs the envelope of the local maxima (upper envelope) and the local minima (lower envelope). The average of these upper and lower envelopes is calculated to obtain the signal's local mean function. This local mean function is subtracted from the original signal to obtain a one-dimensional first-order component, an IMF. A similar operation is then performed on this IMF to obtain a second IMF. This process is iterated until the remaining components become residuals or monotonic functions that approach zero. Adding these IMFs together reconstructs the original signal. Process of EMD decomposition algorithm Figure 4 As shown in Figure 1, the IMF must satisfy the following two conditions: (1) the number of local extreme points and zero crossings in the function must be equal or differ by at most one over the entire time range; (2) at any point in time, the average of the upper and lower envelopes must be zero. In practice, the mean of the upper and lower envelopes cannot be zero. Usually, when the following equation is satisfied, the mean of the envelope is considered to satisfy the IMF's zero mean condition 1. Where h k (t) is the kth IMF, and ε is the screening threshold, which is generally between 0.2 and 0.3.

[0129] In order to reduce the difficulty of model training, in an optional implementation manner, the second input unit includes:

[0130] a first calculation module, performing a skip connection on the hidden layer, transferring the first output information of the first hidden layer to the second hidden layer for calculation, and obtaining the second output information of the second hidden layer, wherein the first hidden layer is the hidden layer currently being calculated, the second hidden layer is the hidden layer not adjacent to the first hidden layer, and the first output information is the output of the first hidden layer and the input of the second hidden layer;

[0131] The second calculation module adds the first output information and the second output information to obtain a sample line loss rate component corresponding to each of the sample modal components.

[0132] In the above-mentioned embodiments, neural networks generally consist of an input layer, hidden layers, and an output layer. Layers are often fully connected, meaning that every neuron in each layer is connected to all neurons in the next layer, while neurons within a layer are not connected to each other. However, due to limitations in the model structure, this model not only fails to effectively learn time series but also suffers from vanishing gradients and other training difficulties during training. Therefore, to address these issues and improve the accuracy of predicting line loss rates in distribution networks containing DGs, a network optimization based on the traditional BP neural network was performed, introducing skip connections. Figure 5 This is a schematic diagram of the structure of the line loss rate determination model. In order to alleviate the gradient disappearance, the output information F(x1) of hidden layer 1 is directly transmitted to hidden layer n, and a jump connection is performed to ensure that the original information is not lost due to nonlinear transformation. Finally, the output layer information F(xn) is added to F(x1) to output the expected data. For example, for any modal component, through the network jump connection, the output vector y of hidden layer 1 is p1 , pass hidden layer 1 directly to hidden layer 3, the output vector of hidden layer 3 is y p2 , so the final output vector y of this modal component is p y p =y p1 +y p2 ,in, x i is the i-th modal component, w i represents the feature weight, and n represents the total number of modal components.

[0133] In order to avoid abnormal line loss rate of the distribution network, in an optional embodiment, the device further includes:

[0134] The monitoring unit is used to, after outputting the target line loss rate of the distribution network, continuously monitor the operation of the distribution network and update the line loss rate determination model periodically or in real time when the target line loss rate is less than or equal to the line loss rate threshold.

[0135] In the above embodiment, when the above target line loss rate is less than or equal to the above line loss rate threshold, it means that the current line loss rate of the distribution network is normal, but it is still necessary to continuously monitor the operation of the above distribution network to prevent abnormal conditions in the distribution network, and it is also necessary to update the above line loss rate determination model regularly or in real time, continuously optimize the operation strategy and control parameters of the distribution network, improve the operation efficiency and performance of the system, reduce energy waste, and reduce operating costs.

[0136] The apparatus for determining a distribution network line loss rate includes a processor and memory. The acquisition unit, input / output unit, and execution unit are all stored as program units in the memory. The processor executes the program units stored in the memory to implement the corresponding functions. All of the modules are located in the same processor; alternatively, the modules can be located in different processors in any combination.

[0137] The processor includes a core, which retrieves the corresponding program unit from the memory. One or more cores can be set, and the problem of inaccurate distribution network line loss rate prediction in the existing technology can be solved by adjusting the core parameters.

[0138] 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, and the memory includes at least one memory chip.

[0139] An embodiment of the present invention provides a computer-readable storage medium, which includes a stored program. When the program is run, the device where the computer-readable storage medium is located is controlled to execute the method for determining the line loss rate of the distribution network.

[0140] Specifically, the method for determining the line loss rate of the distribution network includes:

[0141] Step S201: Acquire a plurality of important line loss indicator data, wherein the important line loss indicator data is indicator data that affects the line loss rate of the distribution network;

[0142] Step S202: Input the plurality of important line loss indicator data into a line loss rate determination model, perform empirical mode decomposition on the important line loss indicator data using the line loss rate determination model to obtain a plurality of time series modal components, and superimpose and reconstruct the line loss rate components corresponding to each of the time series modal components to output a target line loss rate for the distribution network. The line loss rate determination model is a convergence model obtained by performing modal decomposition on a first sample training set and then inputting it into a neural network for training. The first sample training set includes a plurality of groups of first training data, each group of the first training data including sample important indicator data and a first sample line loss rate corresponding to the sample important indicator data.

[0143] Step S203: When the target line loss rate is greater than the line loss rate threshold, a loss reduction operation is performed, where the loss reduction operation is an operation of reducing the line loss rate of the distribution network.

[0144] An embodiment of the present invention provides a processor, which is used to run a program, wherein the method for determining the line loss rate of the distribution network is executed when the program is run.

[0145] Specifically, the method for determining the line loss rate of the distribution network includes:

[0146] Step S201: Acquire a plurality of important line loss indicator data, wherein the important line loss indicator data is indicator data that affects the line loss rate of the distribution network;

[0147] Step S202: Input the plurality of important line loss indicator data into a line loss rate determination model, perform empirical mode decomposition on the important line loss indicator data using the line loss rate determination model to obtain a plurality of time series modal components, and superimpose and reconstruct the line loss rate components corresponding to each of the time series modal components to output a target line loss rate for the distribution network. The line loss rate determination model is a convergence model obtained by performing modal decomposition on a first sample training set and then inputting it into a neural network for training. The first sample training set includes a plurality of groups of first training data, each group of the first training data including sample important indicator data and a first sample line loss rate corresponding to the sample important indicator data.

[0148] Step S203: When the target line loss rate is greater than the line loss rate threshold, a loss reduction operation is performed, where the loss reduction operation is an operation of reducing the line loss rate of the distribution network.

[0149] An embodiment of the present invention provides a system for determining a line loss rate in a distribution network. The system includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, at least the following steps are implemented:

[0150] Step S201: Acquire a plurality of important line loss indicator data, wherein the important line loss indicator data is indicator data that affects the line loss rate of the distribution network;

[0151] Step S202: Input the plurality of important line loss indicator data into a line loss rate determination model, perform empirical mode decomposition on the important line loss indicator data using the line loss rate determination model to obtain a plurality of time series modal components, and superimpose and reconstruct the line loss rate components corresponding to each of the time series modal components to output a target line loss rate for the distribution network. The line loss rate determination model is a convergence model obtained by performing modal decomposition on a first sample training set and then inputting it into a neural network for training. The first sample training set includes a plurality of groups of first training data, each group of the first training data including sample important indicator data and a first sample line loss rate corresponding to the sample important indicator data.

[0152] Step S203: When the target line loss rate is greater than the line loss rate threshold, a loss reduction operation is performed, where the loss reduction operation is an operation of reducing the line loss rate of the distribution network.

[0153] The present application also provides a computer program product, which, when executed on a data processing device, is adapted to execute a program for initializing at least the following method steps:

[0154] Step S201: Acquire a plurality of important line loss indicator data, wherein the important line loss indicator data is indicator data that affects the line loss rate of the distribution network;

[0155] Step S202: Input the plurality of important line loss indicator data into a line loss rate determination model, perform empirical mode decomposition on the important line loss indicator data using the line loss rate determination model to obtain a plurality of time series modal components, and superimpose and reconstruct the line loss rate components corresponding to each of the time series modal components to output a target line loss rate for the distribution network. The line loss rate determination model is a convergence model obtained by performing modal decomposition on a first sample training set and then inputting it into a neural network for training. The first sample training set includes a plurality of groups of first training data, each group of the first training data including sample important indicator data and a first sample line loss rate corresponding to the sample important indicator data.

[0156] Step S203: When the target line loss rate is greater than the line loss rate threshold, a loss reduction operation is performed, where the loss reduction operation is an operation of reducing the line loss rate of the distribution network.

[0157] Obviously, those skilled in the art will appreciate that the various modules or steps of the present invention described above can be implemented using a general-purpose computing device, can be centralized on a single computing device, or can be distributed across a network of multiple computing devices. They can be implemented using program code executable by the computing device, and thus, can be stored in a storage device and executed by the computing device. In some cases, the steps shown or described herein can be performed in a different order than that shown, or can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0158] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0159] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0160] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0161] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0162] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0163] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0164] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0165] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0166] From the above description, it can be seen that the above embodiments of the present application achieve the following technical effects:

[0167] 1) The method for determining the line loss rate of a distribution network of the present application, first, obtains a plurality of important line loss indicator data, the above-mentioned important line loss indicator data being indicator data affecting the line loss rate of the distribution network; then, the plurality of the above-mentioned important line loss indicator data are input into a line loss rate determination model, so as to use the above-mentioned line loss rate determination model to perform empirical mode decomposition on the above-mentioned important line loss indicator data to obtain a plurality of time series modal components, and the line loss rate components corresponding to each of the above-mentioned time series modal components are superimposed and reconstructed, and the target line loss rate of the above-mentioned distribution network is output. The above-mentioned line loss rate determination model is a convergence model obtained by performing modal decomposition on a first sample training set and inputting it into a neural network for training. The above-mentioned first sample training set includes multiple groups of first training data, and each group of the above-mentioned first training data includes sample important indicator data and a first sample line loss rate corresponding to the above-mentioned sample important indicator data; finally, when the above-mentioned target line loss rate is greater than the line loss rate threshold, a loss reduction operation is performed, and the above-mentioned loss reduction operation is an operation to reduce the above-mentioned line loss rate of the above-mentioned distribution network. This application selects multiple important line loss indicator data that have a significant impact on the line loss rate and uses them as input data for the line loss rate determination model. It then uses the empirical mode decomposition method to decompose each important line loss indicator data into multiple modal components. A neural network is then used to predict the line loss rate for each modal component of the important line loss indicator data, obtaining the line loss rate component for each indicator data. All line loss rate components are then superimposed and reconstructed to obtain the target line loss rate. This allows for distribution network line loss rate prediction that takes into account the timing characteristics of DGs, enabling more flexible adaptation to changes in the DG system and making the prediction results more accurate and reliable. This application addresses the issue of inaccurate distribution network line loss rate prediction in the prior art.

[0168] 2) The device for determining the line loss rate of a distribution network of the present application includes an acquisition unit for acquiring a plurality of important line loss indicator data, wherein the above-mentioned important line loss indicator data are indicator data that affect the line loss rate of the distribution network; an input-output unit for inputting the above-mentioned important line loss indicator data into a line loss rate determination model, so as to use the above-mentioned line loss rate determination model to perform empirical mode decomposition on the above-mentioned important line loss indicator data to obtain a plurality of time series modal components, and superimpose and reconstruct the line loss rate components corresponding to each of the above-mentioned time series modal components, and output the target line loss rate of the above-mentioned distribution network. The above-mentioned line loss rate determination model is a convergence model obtained by performing modal decomposition on a first sample training set and inputting it into a neural network for training. The above-mentioned first sample training set includes multiple groups of first training data, and each group of the above-mentioned first training data includes sample important indicator data and a first sample line loss rate corresponding to the above-mentioned important sample indicator data; an execution unit for executing a loss reduction operation when the above-mentioned target line loss rate is greater than the line loss rate threshold. The above-mentioned loss reduction operation is an operation for reducing the above-mentioned line loss rate of the above-mentioned distribution network. This application selects multiple important line loss indicator data that have a significant impact on the line loss rate and uses them as input data for the line loss rate determination model. It then uses the empirical mode decomposition method to decompose each important line loss indicator data into multiple modal components. A neural network is then used to predict the line loss rate for each modal component of the important line loss indicator data, obtaining the line loss rate component for each indicator data. All line loss rate components are then superimposed and reconstructed to obtain the target line loss rate. This allows for distribution network line loss rate prediction that takes into account the timing characteristics of DGs, enabling more flexible adaptation to changes in the DG system and making the prediction results more accurate and reliable. This application addresses the issue of inaccurate distribution network line loss rate prediction in the prior art.

[0169] The above description is merely a preferred embodiment of the present application and is not intended to limit the present application. Various modifications and variations are possible for those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present application shall be included within the scope of protection of the present application.

Claims

1. A method for determining the line loss rate of a distribution network, characterized in that: include: Acquire a plurality of important line loss indicator data, wherein the important line loss indicator data is indicator data affecting the line loss rate of the distribution network; Inputting the plurality of important line loss indicator data into a line loss rate determination model, performing empirical mode decomposition on the important line loss indicator data using the line loss rate determination model to obtain a plurality of time series modal components, and superimposing and reconstructing the line loss rate components corresponding to each of the time series modal components to output a target line loss rate for the distribution network. The line loss rate determination model is a convergence model obtained by performing modal decomposition on a first sample training set and then inputting it into a neural network for training. The first sample training set includes a plurality of groups of first training data, each group of the first training data including sample important indicator data and a first sample line loss rate corresponding to the sample important indicator data. When the target line loss rate is greater than a line loss rate threshold, performing a loss reduction operation, wherein the loss reduction operation is an operation of reducing the line loss rate of the distribution network; Before inputting the plurality of said important line loss index data into the line loss rate determination model, the method further comprises: constructing an initial line loss rate determination model, the said initial line loss rate determination model comprising a modal decomposition module and a neural network module, the said modal decomposition module being used to perform modal decomposition on the input data, the said neural network module being used to calculate the modal decomposed data to obtain output data; a first input step, inputting the said sample important index data in a plurality of groups of current first training data into the said initial line loss rate determination model, the said current first training data being any group of the first training data in the first sample training set; a decomposition step, performing empirical mode decomposition on the said sample important index data to obtain a plurality of sample modal components; a second input step, inputting the said plurality of said The sample modal components are input from the input layer of the neural network, and are calculated through multiple hidden layers of the neural network to obtain sample line loss rate components corresponding to each of the sample modal components; in the reconstruction step, the sample line loss rate components corresponding to each of the sample modal components are superimposed and reconstructed to obtain the current second sample line loss rate; the first input step, the decomposition step, the second input step and the reconstruction step are repeated at least once in sequence until the number of iterations reaches the maximum number of iterations or the second loss value of the second loss function is less than the second loss threshold, to obtain the line loss rate determination model, the second loss function is the mean square error function between the first sample line loss rate and the current second sample line loss rate, and the second loss value is the value of the mean square error function.

2. The method according to claim 1, characterized in that Obtain multiple important line loss indicator data, including: Acquire all indicator data of the distribution network, where the indicator data is data corresponding to various indicator types, and the indicator types include at least feeder power supply radius, daily load rate fluctuation coefficient, maximum temperature, minimum temperature, and weather type; Performing noise reduction processing on all the indicator data to obtain first indicator data corresponding to each indicator type; Normalizing all the first indicator data to obtain second indicator data corresponding to each indicator type; The second indicator data are sequentially input into a linear regression model to obtain a feature weight corresponding to each second indicator data. The linear regression model is a convergent model obtained by inputting the second sample training set into a linear regression equation for training. The second sample training set includes multiple groups of second training data, each group of second training data includes sample second indicator data, and a second sample line loss rate and a sample feature weight corresponding to the sample second indicator data. Sort all the feature weights from large to small to obtain a weight sequence; The second indicator data corresponding to the feature weights whose sequence numbers in the weight sequence are less than or equal to a set number are selected and determined as the important line loss indicator data.

3. The method according to claim 2, characterized in that Performing noise reduction processing on all the indicator data to obtain first indicator data corresponding to each indicator type includes: Removing all null values ​​from the indicator data to obtain first noise reduction indicator data corresponding to each indicator type; The deviation of each indicator type corresponding to the first noise reduction indicator data is calculated according to the first formula. The first formula is: , the determination of the distribution network line loss rate represents the deviation degree, i represents the number of the first noise reduction index data, represents the i-th first noise reduction indicator data corresponding to the current indicator type, represents the mean value of all the first noise reduction index data corresponding to the current index type, Indicates the standard deviation of the first noise reduction index data corresponding to the current index type; When the absolute value of the deviation is greater than or equal to the deviation threshold, the first noise reduction index data is eliminated; When the absolute value of the deviation is smaller than the deviation threshold, the first noise reduction index data is determined as the first index data.

4. The method according to claim 2, characterized in that Before sequentially inputting the second indicator data into the linear regression model, the method further includes: Constructing an initial linear regression model, wherein the initial linear regression model is a linear regression model to be trained; an analysis step of inputting the sample second indicator data from multiple sets of current second training data into the initial linear regression model for linear regression analysis to obtain the corresponding current first sample line loss rate, wherein the current second training data is any set of the second training data in the second sample training set; an optimization step, using a mean square error as a first loss function and optimizing the first loss function using a least squares method, wherein the first loss function is used to measure a difference between the current first sample line loss rate and the second sample line loss rate of the sample second indicator data; Repeat the analysis step and the optimization step in sequence until the first loss value of the first loss function reaches the minimum, obtain the linear regression model, and determine the regression coefficient of the linear regression model as the sample feature weight, and the first loss value is the difference between the current first sample line loss rate and the second sample line loss rate of the sample second indicator data.

5. The method according to claim 1, wherein Inputting a plurality of the sample modal components from the input layer of the neural network and performing calculations through a plurality of hidden layers of the neural network to obtain a sample line loss rate component corresponding to each of the sample modal components includes: Performing a skip connection on the hidden layer, passing the first output information of the first hidden layer to the second hidden layer for calculation, and obtaining the second output information of the second hidden layer, wherein the first hidden layer is the hidden layer currently being calculated, the second hidden layer is the hidden layer not adjacent to the first hidden layer, and the first output information is the output of the first hidden layer and the input of the second hidden layer; The first output information and the second output information are added together to obtain a sample line loss rate component corresponding to each of the sample modal components.

6. The method according to claim 1, wherein After outputting the target line loss rate of the distribution network, the method further includes: When the target line loss rate is less than or equal to the line loss rate threshold, the operation of the distribution network is continuously monitored, and the line loss rate determination model is updated periodically or in real time.

7. A device for determining line loss rate of a distribution network, characterized in that: The device comprises: An acquisition unit, configured to acquire a plurality of important line loss indicator data, wherein the important line loss indicator data is indicator data affecting the line loss rate of the distribution network; an input / output unit, configured to input the plurality of important line loss indicator data into a line loss rate determination model, perform empirical mode decomposition on the important line loss indicator data using the line loss rate determination model to obtain a plurality of time series modal components, and superimpose and reconstruct the line loss rate components corresponding to the time series modal components to output a target line loss rate for the distribution network, wherein the line loss rate determination model is a convergence model obtained by performing modal decomposition on a first sample training set and then inputting the resultant data into a neural network for training, wherein the first sample training set includes a plurality of groups of first training data, each group of the first training data including sample important indicator data and a first sample line loss rate corresponding to the sample important indicator data; an execution unit, configured to execute a loss reduction operation when the target line loss rate is greater than a line loss rate threshold, wherein the loss reduction operation is an operation of reducing the line loss rate of the distribution network; The device further includes: a second construction unit for constructing an initial line loss rate determination model before inputting a plurality of the line loss important indicator data into the line loss rate determination model, the initial line loss rate determination model including a modal decomposition module and a neural network module, the modal decomposition module for performing modal decomposition on the input data, and the neural network module for calculating the modal decomposed data to obtain output data; a first input unit for executing a first input step, inputting the sample important indicator data in a plurality of groups of current first training data into the initial line loss rate determination model, the current first training data being any group of the first training data in the first sample training set; a decomposition unit for executing a decomposition step, performing empirical mode decomposition on the sample important indicator data to obtain a plurality of sample modal components; a second input unit for executing a first input step, The second input step inputs the plurality of sample modal components from the input layer of the neural network, calculates through the plurality of hidden layers of the neural network, and obtains the sample line loss rate components corresponding to each of the sample modal components; the reconstruction unit is used to execute the reconstruction step, superimpose and reconstruct the sample line loss rate components corresponding to each of the sample modal components, and obtain the current second sample line loss rate; the second repetition unit is used to repeat the first input step, the decomposition step, the second input step, and the reconstruction step at least once in sequence until the number of iterations reaches the maximum number of iterations or the second loss value of the second loss function is less than the second loss threshold, and obtain the line loss rate determination model, the second loss function is the mean square error function between the first sample line loss rate and the current second sample line loss rate, and the second loss value is the value of the mean square error function.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the method according to any one of claims 1 to 6.

9. A computer program product comprising computer instructions, characterized in that When the computer instructions are executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images