Determination method and device of power line loss interval determination model, equipment and storage medium

By mapping power data to three-dimensional space and using convolutional neural network for training, the problem of low model training efficiency in power line loss calculation is solved, and the accuracy and reliability of the line loss interval determination model is improved.

CN119004977APending Publication Date: 2024-11-22GUANGDONG POWER GRID CO LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411082451.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-08
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

The existing power line loss calculation methods have the problem of low model training efficiency, which affects the accuracy and reliability of the power line loss interval determination model.

Method used

By collecting historical and current power data, map it to three-dimensional space to generate power mimetic diagrams, and use convolutional neural networks for offline and online training to reduce the impact of abnormal data noise and improve model training efficiency and performance.

Benefits of technology

The training efficiency and performance of the power line loss interval determination model is improved, the problem of low model training efficiency is solved, and more accurate line loss interval prediction and management is achieved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119004977A_ABST
    Figure CN119004977A_ABST
Patent Text Reader

Abstract

The invention discloses a determination method and device of a power line loss interval determination model, equipment and a storage medium. The method comprises the steps that a preset number of electric power data are collected, and the electric power data comprise historical electric power data and current electric power data; mapping the historical power data to a three-dimensional space to obtain a power imitation graph, and performing offline training on a pre-established convolutional neural network model based on the historical power data and the power imitation graph to obtain an initial power line loss interval determination model; and performing online training on the initial power line loss interval determination model based on the current power data to obtain a target power line loss interval determination model. And the training efficiency and the model performance of the power line loss interval determination model are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of power line loss, and in particular to a method, device, equipment and storage medium for determining a power line loss interval determination model. Background Art

[0002] Line loss is an important indicator of lean management and energy saving and loss reduction of power equipment. It can reflect the energy management level of a power supply company. At present, power grid companies adopt a "one-size-fits-all" line loss assessment standard, that is, regardless of the specific operation of the power equipment area, a unified standard is adopted. This has led to overly strict or conservative inspections in some power equipment areas, which is not conducive to the economic operation of the power equipment area, and has brought difficulties to the line loss management of power equipment. Therefore, the research on line loss calculation and loss reduction measures of power equipment has been paid more and more attention.

[0003] In recent years, based on the traditional method of calculating power loss, the line loss calculation method of each branch line can be obtained by using the field monitoring terminal. In addition, a network topology identification method based on fuzzy clustering was introduced, and a power grid state assessment model based on direct neural dynamic programming was established, which achieved good results in theoretical line loss calculation. With the popularization of data acquisition terminals and the development of information technology, more and more studies are applying big data technology and artificial intelligence methods to distribution network line loss calculation in a data-driven manner. Using various neural network deep learning frameworks, the power supply, power factor, total harmonic distortion, unbalance and load shape factor are used as inputs, and the predicted line loss rate is output. However, both the traditional impedance-based line loss calculation method and the line loss prediction and division method based on intelligent algorithms have certain defects. In the process of power production, abnormal measurement equipment and abnormal data transmission lead to problems in the collected data, which in turn leads to low model training efficiency. Summary of the invention

[0004] The present invention provides a method, device, equipment and storage medium for determining air intake amount based on an exhaust gas recirculation system, so as to solve the problem of high engine cost caused by setting up a flow sensor at the front end of the intake system.

[0005] According to one aspect of the present invention, a method for determining a power line loss interval determination model is provided, the method comprising:

[0006] Collecting a preset amount of power data, wherein the power data includes historical power data and current power data;

[0007] Mapping the historical power data to a three-dimensional space to obtain a power simulation map, and performing offline training on a pre-established convolutional neural network model based on the historical power data and the power simulation map to obtain an initial power line loss interval determination model;

[0008] The initial power line loss interval determination model is trained online based on the current power data to obtain a target power line loss interval determination model.

[0009] According to another aspect of the present invention, there is provided a device for determining a power line loss interval determination model, the device comprising:

[0010] A data collection module, used to collect a preset amount of power data, wherein the power data includes historical power data and current power data;

[0011] A simulation graph training module, used for mapping the historical power data into a three-dimensional space to obtain a power simulation graph, and performing offline training on a pre-established convolutional neural network model based on the historical power data and the power simulation graph to obtain an initial power line loss interval determination model;

[0012] The model determination module is used to perform online training on the initial power line loss interval determination model based on the current power data to obtain a target power line loss interval determination model.

[0013] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0014] at least one processor; and

[0015] a memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the method for determining the power line loss interval determination model described in any embodiment of the present invention.

[0017] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the method for determining the power line loss interval determination model described in any embodiment of the present invention when executed.

[0018] The technical solution of the embodiment of the present invention collects a preset amount of power data, wherein the power data includes historical power data and current power data; to obtain a variety of power data; then, maps the historical power data to a three-dimensional space to obtain a power simulation map, and performs offline training on a pre-established convolutional neural network model based on the historical power data and the power simulation map to obtain an initial power line loss interval determination model; maps time domain information to spatial information to reduce the impact of noise generated by abnormal data on model training; finally, performs online training on the initial power line loss interval determination model based on the current power data to obtain a target power line loss interval determination model, thereby solving the problem of low model training efficiency and achieving the beneficial effect of improving the training efficiency and model performance of the power line loss interval determination model.

[0019] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 is a flow chart of a method for determining a power line loss interval determination model provided according to Embodiment 1 of the present invention;

[0022] Figure 2a is a flow chart of a method for determining a power line loss interval determination model provided according to Embodiment 2 of the present invention;

[0023] Figure 2b is a schematic structural diagram of an initial convolutional neural network of an optional example of a method for determining a power line loss interval determination model provided in Embodiment 2 of the present invention;

[0024] Figure 2c It is a flowchart of line loss assessment and model online training of an optional example of a method for determining a power line loss interval determination model provided in accordance with the second embodiment of the present invention;

[0025] Figure 3 is a schematic structural diagram of a device for determining a power line loss interval determination model provided according to a third embodiment of the present invention;

[0026] Figure 4It is a structural schematic diagram of an electronic device for implementing a method for determining a power line loss interval determination model according to an embodiment of the present invention. DETAILED DESCRIPTION

[0027] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0028] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0029] Embodiment 1

[0030] Figure 1 A flowchart of a method for determining a power line loss interval determination model is provided for the first embodiment of the present invention. This embodiment is applicable to the case of power line loss analysis. The method can be executed by a device for determining a power line loss interval determination model. The device for determining a power line loss interval determination model can be implemented in the form of hardware and / or software. The device for determining a power line loss interval determination model can be configured in an electronic device. Figure 1 As shown, the method includes:

[0031] S110. Collect a preset amount of power data, wherein the power data includes historical power data and current power data.

[0032] The power data may be understood as the operating data of a power equipment area, for example, it may be the monthly operating data of a single power equipment area.

[0033] Specifically, a preset amount of current power data and a preset amount of historical data are collected through the power consumption information collection system, for example, current monthly (30 days) power data and historical monthly power data are collected.

[0034] S120, mapping the historical power data to a three-dimensional space to obtain a power simulation map, and performing offline training on a pre-established convolutional neural network model based on the historical power data and the power simulation map to obtain an initial power line loss interval determination model.

[0035] Specifically, the collected two-dimensional historical power data is mapped to a three-dimensional space to obtain a power simulation map in a simulation map format. The pre-established convolutional neural network model is trained offline iteratively through the historical power data and the power simulation map to obtain an initial power line loss interval determination model completed by offline training.

[0036] Optionally, the convolutional neural network model includes a first input layer, a first convolutional neural network connected to the first input layer, a second input layer, a second convolutional neural network connected to the second input layer, a splicing layer connected to the first convolutional neural network and the second convolutional neural network respectively, a classifier connected to the splicing layer, and an output layer connected to the classifier, the first convolutional neural network includes multiple convolutional layers, a batch normalization layer and a global average pooling layer, and the second convolutional neural network includes a one-hot encoding layer and a fully connected layer connected to the one-hot encoding layer.

[0037] Optionally, the pre-established convolutional neural network model is offline trained based on the historical power data and the power simulation graph, including: inputting the power simulation graph into the first convolutional neural network through the first input layer, extracting the image feature vector through the convolution layer of the first convolutional neural network, regularizing the image feature vector through the batch normalization layer, and extracting the first feature vector from the regularized image feature vector through the global average pooling layer; inputting the historical power data into the second convolutional neural network through the second input layer, encoding the historical power data into a sparse feature vector through the one-hot encoding layer, and extracting the second feature vector from the sparse feature vector through the fully connected layer; splicing the first feature vector and the second feature vector through the splicing layer to obtain a target feature vector, classifying the target feature vector into a line loss interval through the classifier, outputting the line loss interval through the output layer, and adjusting the model parameters of the convolutional neural network model based on the line loss interval output by the output layer.

[0038] Specifically, the power simulation image is sent as input data to the first convolutional neural network through the first input layer, and the feature vector in the image is extracted through the convolution layer of the first convolutional neural network. The extracted image feature vector is regularized through the batch normalization layer to reduce the internal covariate shift and improve the training speed and stability of the model. The first feature vector is extracted from the regularized image feature vector through the global average pooling layer for subsequent feature splicing. The historical power data is sent to the second convolutional neural network through the second input layer, and the historical power data is encoded into a sparse feature vector through the one-hot encoding layer for feature extraction. The second feature vector is extracted from the sparse feature vector through the fully connected layer. The first feature vector and the second feature vector are spliced ​​through the splicing layer to obtain a target feature vector containing image and historical data information, and the target feature vector is classified into line loss intervals through the classifier, and the classification results are output through the output layer, and the model parameters of the convolutional neural network model are adjusted according to the output results to optimize the classification performance of the model.

[0039] In the embodiment of the present invention, the model combines image data and historical power data, and extracts and classifies features through deep learning methods to achieve accurate prediction of power line loss intervals. Different types of data information are fully utilized to improve the accuracy and reliability of line loss interval classification.

[0040] Optionally, the image feature vector is regularized by the batch normalization layer, which is expressed as:

[0041]

[0042] c i =γy' i +β

[0043] Among them, y i is the i-th feature data to be processed in the original batch, and there are n feature data to be processed in the original batch; μ B is the mean of the feature data to be processed in each batch, σ 2 is the variance of the feature data to be processed for each batch, y' k is the regularized y i , ω is the correction amount; c i is the batch normalization output, γ and β are used to correct c i Variable quantity.

[0044] In an embodiment of the present invention, a batch normalization layer is added to regularize each input feature, thereby reducing gradient loss, accelerating network convergence, simplifying tuning participation, increasing network stability, and preventing overfitting.

[0045] Optionally, before offline training the pre-established convolutional neural network model based on the historical power data and the power simulation graph data, the method further includes:

[0046] The historical power data is normalized and expressed as follows:

[0047]

[0048] in, is the normalized data, y(n) is the historical power data, is the minimum value in the historical power data. This is the maximum value in historical electricity data.

[0049] It is understandable that for numerical data such as voltage, current, and power factor, there is a clear numerical relationship between them. If the original data is used directly for training, the model training effect may be poor due to the large difference in the numerical range. By normalizing the historical power data, the data can be scaled to a smaller numerical range, such as between 0 and 1, or between -1 and 1, which makes it easier for the model to learn the intrinsic relationship between the data and improve the training efficiency and accuracy of the model. After normalization, these data can be used as input for neural network training to ensure that the model can correctly classify the line loss range.

[0050] S130: Perform online training on the initial power line loss interval determination model based on the current power data to obtain a target power line loss interval determination model.

[0051] Specifically, after offline training of the pre-built convolutional neural network model to obtain the initial power line loss interval determination model, the initial power line loss interval determination model is continuously trained online through the current power data to obtain a target power line loss interval determination model with higher accuracy.

[0052] The technical solution of the embodiment of the present invention collects a preset amount of power data, wherein the power data includes historical power data and current power data; to obtain a variety of power data; then, maps the historical power data to a three-dimensional space to obtain a power simulation map, and performs offline training on a pre-established convolutional neural network model based on the historical power data and the power simulation map to obtain an initial power line loss interval determination model; maps time domain information to spatial information to reduce the impact of noise generated by abnormal data on model training; finally, performs online training on the initial power line loss interval determination model based on the current power data to obtain a target power line loss interval determination model, thereby solving the problem of low model training efficiency and achieving the beneficial effect of improving the training efficiency and model performance of the power line loss interval determination model.

[0053] Embodiment 2

[0054] Figure 2a A flowchart of a method for determining a power line loss interval determination model provided in Embodiment 2 of the present invention, this embodiment is a further refinement of how to perform online training on the initial power line loss interval determination model based on the current power data in the above embodiment to obtain a target power line loss interval determination model. Optionally, the online training of the initial power line loss interval determination model based on the current power data to obtain a target power line loss interval determination model includes: dividing the current power data into a training set and a test set; performing online iterative training on the initial line loss interval determination model based on the training set, and evaluating the initial line loss interval determination model that completes each training based on the test set to obtain a model performance indicator corresponding to the initial line loss interval determination model; and determining the target line loss interval determination model based on at least one model performance indicator.

[0055] like Figure 2a As shown, the method includes:

[0056] S210. Collect a preset amount of power data, wherein the power data includes historical power data and current power data.

[0057] S220, mapping the historical power data to a three-dimensional space to obtain a power simulation map, and performing offline training on a pre-established convolutional neural network model based on the historical power data and the power simulation map to obtain an initial power line loss interval determination model.

[0058] Optionally, the historical power data includes historical voltage data, historical current data, historical power factor data, data on the number of collection points, data on collection date, and data on collection time series; and mapping the historical power data to a three-dimensional space to obtain a power simulation map includes:

[0059] The historical voltage data, the historical current data, the historical power factor data and the collected time series data are respectively mapped to a three-dimensional space to obtain power simulation map data, wherein the power simulation map includes a voltage channel, a current channel and a power factor channel, the height of the power simulation map is determined based on the collection date data, and the width of the power simulation map is determined based on the collection point quantity data.

[0060] Specifically, the power simulation diagram includes a voltage channel, a current channel, and a power factor channel. Each channel corresponds to a specific physical quantity (voltage, current, power factor) and will be represented in a three-dimensional space. The historical voltage data is mapped to the voltage channel to represent the change in historical voltage in a three-dimensional space. The historical current data is mapped to the current channel to represent the change in historical current in a three-dimensional space. The historical power factor data is mapped to the power factor channel to represent the change in historical power factor in a three-dimensional space. The height of the power simulation diagram is determined based on the collection date data. The height can be set according to the date range or a specific time interval. The width of the power simulation diagram is determined based on the number of collection point data. Each collection point can have a corresponding width unit in the simulation diagram.

[0061] Exemplarily, for the monthly operation data of a single power equipment area, taking the voltage data of 30 days per month as an example, there will be 24×30 collection points, each of which has its specific time attributes and physical meaning; the image-simulating sampling method is used to map the data of the 24×30 voltage, current, and power factor collection points to three-dimensional space to form a 3×30×24 tensor; the tensor has 3 channels, 28, 30 or 31 high and 24 wide; the height represents the collection date of each month, which is determined by the number of days in the month; the 24 points of width represent the 24 collection points of the day; the three channels represent voltage, current, and power factor, respectively; information of 24 time periods in a day is collected, and the time series information is mapped from left to right to spatial position information; 28, 30, and 31 points of equal width represent the data changes of each day in the same time period. Points of the same width and height on different channels represent the voltage, current and power factor information at that time respectively. Through the above method, the voltage, current and power factor information containing 24×28 / 30 / 31 acquisition points in a month can be mapped into an image-like sample format.

[0062] In the embodiment of the present invention, historical voltage data, historical current data, historical power factor data and acquisition time series data are effectively mapped to three-dimensional space to generate a power simulation diagram including voltage channel, current channel and power factor channel. Such simulation diagram can provide strong support for the analysis and visualization of the power system.

[0063] S230: Divide the current power data into a training set and a test set.

[0064] Exemplarily, the current monthly power data may be divided into a training set and a test set according to a preset ratio, and the preset ratio is set in advance based on experience, for example, 70% as a training set and 30% as a test set, which is not limited in this embodiment.

[0065] S240, performing online iterative training on the initial line loss interval determination model based on the training set, and evaluating the initial line loss interval determination model that has completed training each time based on the test set to obtain a model performance indicator corresponding to the initial line loss interval determination model.

[0066] Specifically, the training set is used to perform online iterative training on the initial line loss interval determination model. The model uses new training data to update its parameters in each iteration to improve prediction accuracy. After each iteration of training, the trained initial line loss interval determination model is evaluated using the test set. By comparing the classification results output by the model with the actual results, the model performance indicators corresponding to the initial line loss interval determination model are obtained.

[0067] S250. Determine the target line loss interval determination model based on at least one model performance indicator.

[0068] Specifically, the initial line loss interval determination model with the best performance after training is determined based on at least one model performance indicator, and the model with the best performance is selected as the target line loss interval determination model.

[0069] The technical solution of the embodiment of the present invention is to divide the current power data into a training set and a test set; perform online iterative training on the initial line loss interval determination model based on the training set, and evaluate the initial line loss interval determination model that has completed training each time based on the test set to obtain a model performance index corresponding to the initial line loss interval determination model; and determine the target line loss interval determination model based on at least one model performance index. This achieves the effective determination of a target line loss interval determination model, which can accurately predict the line loss interval of the power system based on the current power data, and provide strong support for the optimization and operation of the power system.

[0070] Figure 2b A schematic diagram of the structure of an initial convolutional neural network of an optional example of a method for determining a power line loss interval determination model is provided. Figure 2b As shown, the initial convolutional neural network includes an upper convolutional neural network, a discard layer, an information concatenation layer, a lower neural network, a one-hot encoding layer, and a fully connected layer. Among them, the upper convolutional neural network is the first convolutional neural network, and the lower convolutional neural network is the second convolutional neural network.

[0071] First, a preset amount of power data is collected, wherein the power data includes historical power data and current power data; the historical power data is mapped to a three-dimensional space to obtain a power simulation map. Exemplarily, the basic data related to voltage, current, power factor, monthly line loss and billiard area for one month are collected through the power consumption information collection system, including information for 24 time periods every day, and the time series information is mapped from left to right to the spatial position. 28, 30, and 31 points of equal width represent the data changes for each day in the same time period. Points of the same width and height on different channels represent the voltage, current, and power factor information at that time. In this way, the voltage, current, and power factor information containing 24×28 / 30 / 31 collection points in a month are mapped into a sample format of a simulated image while retaining the original time series information.

[0072] Since the voltage, current and power factor data are all numerical data, there is a clear numerical relationship between them; the neural network is sensitive to the input data, so the historical power data is normalized and expressed as follows:

[0073]

[0074] in, is the normalized data, y (n) For historical power data, is the minimum value in the historical power data. This is the maximum value in historical electricity data.

[0075] Next, based on the pre-established convolutional neural network, the model has two layers of input: the upper layer input is the power simulation map, which passes through multiple convolutional layers, batch normalization layers, pooling and discard layers, and then extracts the first eigenvector through the global average pooling layer. The lower layer input is the historical power data, which is encoded into a sparse vector through one-hot encoding, and the second eigenvector is extracted through the fully connected layer. Finally, the first eigenvector and the second eigenvector extracted by the two inputs are spliced ​​together, the target eigenvector is classified into line loss intervals through the classifier, the line loss interval is output through the output layer, and the model parameters of the convolutional neural network model are adjusted based on the line loss interval output by the output layer. By training and verifying the constructed data set, we can obtain the mapping relationship between historical power data and transformer area line loss. After the training is completed, the characteristic operation data such as voltage, current, power factor, etc. of the transformer area are input, and the reasonable line loss interval of the transformer area can be obtained through the model output.

[0076] In addition, adding a batch normalization layer can reduce gradient loss, speed up network convergence, simplify tuning, increase network stability, and prevent overfitting by regularizing each input feature. The image feature vector is regularized by the batch normalization layer, which is expressed as:

[0077]

[0078] c i =γy′ i +β

[0079] Among them, y i is the i-th feature data to be processed in the original batch, and there are n feature data to be processed in the original batch; μ B is the mean of the feature data to be processed in each batch, σ 2 is the variance of the feature data to be processed for each batch, y' k is the regularized y i , ω is the correction amount; c i is the batch normalization output, γ and β are used to correct c i Variable quantity.

[0080] Figure 2c A flowchart of line loss assessment and model online training for an optional example of a method for determining a power line loss interval determination model is provided. Figure 2c As shown, first, the current power data is obtained. Then, according to the "7-3" criterion, the data set is divided, and 70% of it is used as a training set for online training of an accurate line loss range determination model. The remaining 30% of the samples constitute the test set, and the model is used to calculate the reasonable line loss range of the corresponding distribution area.

[0081] Secondly, establish an adaptive line loss assessment system; the power line loss interval evaluation system is divided into 5 intervals, namely, reasonable interval, low loss interval, medium loss interval, high loss interval and ultra-high loss interval; based on the trained target power line loss interval determination model, when the data of any transformer in the test set is input into the target power line loss interval determination model, the target power line loss interval determination model will automatically output the reasonable line loss rate interval of the model; compare the model output results with the actual line loss, and perform differential rectification and loss reduction management; after the power equipment is transformed, the iterative process of data collection, data set division and model training will be repeated next month until the rectification is completed when the line loss level of each transformer area reaches the reasonable line loss range.

[0082] Finally, after the power equipment transformation is completed, the iterative process of power data collection, data set division and model training will continue for the next month. This continuous improvement cycle will continue until the line loss level in each area reaches a reasonable line loss range.

[0083] The technical solution of this embodiment is to first process the collected power data in an image format, and then map the time domain information to the spatial information by making power data samples through graph simulation. Secondly, a reasonable line loss interval calculation model is constructed based on a convolutional neural network, and the reasonable line loss intervals for each month in the transformer area are calculated through graph simulation data. Finally, an adaptive evaluation system is constructed based on the reasonable line loss interval model, and the loss is adaptively evaluated according to the reasonable line loss intervals of different power equipment areas output by the model, forming a loss reduction strategy for abnormal power equipment. The proposed power line loss analysis method with adaptive evaluation of graph-simulated convolutional neural network can solve the abnormalities of measuring equipment in the power production process and the abnormalities in the data transmission process, realize the accurate evaluation of the line loss level in the power equipment area, and improve the computational efficiency of the convolutional neural network and the quality of lean management of line losses.

[0084] Embodiment 3

[0085] Figure 3 This is a schematic diagram of the structure of a device for determining a power line loss interval determination model provided in Embodiment 3 of the present invention. Figure 3 As shown, the device includes: a data acquisition module 310, an image simulation training module 320 and a model determination module 330.

[0086] Among them, the data acquisition module 310 is used to collect a preset amount of power data, wherein the power data includes historical power data and current power data; the simulation graph training module 320 is used to map the historical power data to a three-dimensional space to obtain a power simulation graph, and to perform offline training on a pre-established convolutional neural network model based on the historical power data and the power simulation graph to obtain an initial power line loss interval determination model; the model determination module 330 is used to perform online training on the initial power line loss interval determination model based on the current power data to obtain a target power line loss interval determination model.

[0087] The technical solution of the embodiment of the present invention collects a preset amount of power data, wherein the power data includes historical power data and current power data; to obtain a variety of power data; then, maps the historical power data to a three-dimensional space to obtain a power simulation map, and performs offline training on a pre-established convolutional neural network model based on the historical power data and the power simulation map to obtain an initial power line loss interval determination model; maps time domain information to spatial information to reduce the impact of noise generated by abnormal data on model training; finally, performs online training on the initial power line loss interval determination model based on the current power data to obtain a target power line loss interval determination model, thereby solving the problem of low model training efficiency and achieving the beneficial effect of improving the training efficiency and model performance of the power line loss interval determination model.

[0088] Optionally, the convolutional neural network model includes a first input layer, a first convolutional neural network connected to the first input layer, a second input layer, a second convolutional neural network connected to the second input layer, a splicing layer connected to the first convolutional neural network and the second convolutional neural network respectively, a classifier connected to the splicing layer, and an output layer connected to the classifier, the first convolutional neural network includes multiple convolutional layers, a batch normalization layer and a global average pooling layer, and the second convolutional neural network includes a one-hot encoding layer and a fully connected layer connected to the one-hot encoding layer.

[0089] Optionally, the image imitation training module includes:

[0090] a first feature vector extraction unit, configured to input the power simulation image into a first convolutional neural network through the first input layer, extract an image feature vector through the convolutional layer of the first convolutional neural network, perform regularization processing on the image feature vector through the batch normalization layer, and extract a first feature vector from the regularized image feature vector through the global average pooling layer;

[0091] a second feature vector extraction unit, configured to input the historical power data into a second convolutional neural network through the second input layer, encode the historical power data into a sparse feature vector through a one-hot encoding layer, and extract a second feature vector from the sparse feature vector through the fully connected layer;

[0092] A model parameter adjustment unit is used to splice the first feature vector and the second feature vector through the splicing layer to obtain a target feature vector, classify the target feature vector into line loss intervals through the classifier, output the line loss intervals through the output layer, and adjust the model parameters of the convolutional neural network model based on the line loss intervals output by the output layer.

[0093] Optionally, the image feature vector is regularized by the batch normalization layer, which is expressed as:

[0094]

[0095] c i =γy′ i +β

[0096] Among them, y i is the i-th feature data to be processed in the original batch, and there are n feature data to be processed in the original batch; μ B is the mean of the feature data to be processed in each batch, σ 2 is the variance of the feature data to be processed for each batch, y' k is the regularized y i, ω is the correction amount; c i is the batch normalization output, γ and β are used to correct c i Variable quantity.

[0097] Optionally, the device also includes a normalization processing module.

[0098] The normalization processing module is used to perform normalization processing on the historical power data before the pre-established convolutional neural network model is offline trained based on the historical power data and the power simulation data, and is expressed by the formula:

[0099]

[0100] in, is the normalized data, y (n) For historical power data, is the minimum value in the historical power data. This is the maximum value in historical electricity data.

[0101] Optionally, the historical power data includes historical voltage data, historical current data, historical power factor data, data on the number of collection points, data on collection date, and collection time series data; accordingly, the image simulation training module is specifically used for:

[0102] The historical voltage data, the historical current data, the historical power factor data and the collected time series data are respectively mapped to a three-dimensional space to obtain power simulation map data, wherein the power simulation map includes a voltage channel, a current channel and a power factor channel, the height of the power simulation map is determined based on the collection date data, and the width of the power simulation map is determined based on the collection point quantity data.

[0103] Optionally, the model determination module includes:

[0104] A data set division unit, used for dividing the current power data into a training set and a test set;

[0105] An online training and evaluation unit, configured to perform online iterative training on the initial line loss interval determination model based on the training set, and to evaluate the initial line loss interval determination model that has completed training each time based on the test set, so as to obtain a model performance indicator corresponding to the initial line loss interval determination model;

[0106] A model determination unit is used to determine the target line loss interval determination model based on at least one model performance indicator.

[0107] The device for determining a power line loss interval determination model provided in an embodiment of the present invention can execute the method for determining a power line loss interval determination model provided in any embodiment of the present invention, and has functional modules and beneficial effects corresponding to the execution method.

[0108] Embodiment 4

[0109] Figure 4 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0110] like Figure 4 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0111] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0112] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The processor 11 executes the various methods and processes described above, such as the determination of the power line loss interval determination model.

[0113] In some embodiments, the method of determining the power line loss interval determination model can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of determining the power line loss interval determination model described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to perform the method of determining the power line loss interval determination model in any other appropriate manner (for example, by means of firmware).

[0114] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0115] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0116] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0117] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0118] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0119] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0120] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0121] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for determining a power line loss interval determination model, characterized in that: include: Collecting a preset amount of power data, wherein the power data includes historical power data and current power data; Mapping the historical power data to a three-dimensional space to obtain a power simulation map, and performing offline training on a pre-established convolutional neural network model based on the historical power data and the power simulation map to obtain an initial power line loss interval determination model; The initial power line loss interval determination model is trained online based on the current power data to obtain a target power line loss interval determination model.

2. The method according to claim 1, characterized in that: The convolutional neural network model includes a first input layer, a first convolutional neural network connected to the first input layer, a second input layer, a second convolutional neural network connected to the second input layer, a splicing layer connected to the first convolutional neural network and the second convolutional neural network respectively, a classifier connected to the splicing layer, and an output layer connected to the classifier. The first convolutional neural network includes multiple convolutional layers, a batch normalization layer, and a global average pooling layer. The second convolutional neural network includes a one-hot encoding layer and a fully connected layer connected to the one-hot encoding layer.

3. The method according to claim 2, characterized in that The offline training of the pre-established convolutional neural network model based on the historical power data and the power simulation graph includes: Inputting the power simulation image into a first convolutional neural network through the first input layer, extracting an image feature vector through the convolutional layer of the first convolutional neural network, regularizing the image feature vector through the batch normalization layer, and extracting a first feature vector from the regularized image feature vector through the global average pooling layer; Inputting the historical power data into a second convolutional neural network through the second input layer, encoding the historical power data into a sparse feature vector through a one-hot encoding layer, and extracting a second feature vector from the sparse feature vector through the fully connected layer; The first feature vector and the second feature vector are spliced ​​through the splicing layer to obtain a target feature vector, the target feature vector is classified into line loss intervals through the classifier, the line loss intervals are output through the output layer, and the model parameters of the convolutional neural network model are adjusted based on the line loss intervals output by the output layer.

4. The method according to claim 3, characterized in that The image feature vector is regularized by the batch normalization layer, which is expressed as: c i =γy′ i +b Among them, y i is the i-th feature data to be processed in the original batch, and there are n feature data to be processed in the original batch; μ B is the mean of the feature data to be processed in each batch, σ 2 is the variance of the feature data to be processed for each batch, y' k is the regularized y i , ω is the correction amount; c i is the batch normalization output, γ and β are used to correct c i Variable amount.

5. The method according to claim 1, characterized in that Before offline training of the pre-established convolutional neural network model based on the historical power data and the power simulation graph data, the method further includes: The historical power data is normalized and expressed as follows: in, is the normalized data, y(n) is the historical power data, is the minimum value in the historical power data. This is the maximum value in historical electricity data.

6. The method according to claim 1, characterized in that The historical power data includes historical voltage data, historical current data, historical power factor data, data on the number of collection points, data on collection date, and collection time series data; the historical power data is mapped to a three-dimensional space to obtain a power simulation map, including: The historical voltage data, the historical current data, the historical power factor data and the collected time series data are respectively mapped to a three-dimensional space to obtain power simulation map data, wherein the power simulation map includes a voltage channel, a current channel and a power factor channel, the height of the power simulation map is determined based on the collection date data, and the width of the power simulation map is determined based on the collection point quantity data.

7. The method according to claim 1, characterized in that The online training of the initial power line loss interval determination model based on the current power data to obtain a target power line loss interval determination model includes: Dividing the current power data into a training set and a test set; Performing online iterative training on the initial line loss interval determination model based on the training set, and evaluating the initial line loss interval determination model that has completed training each time based on the test set, so as to obtain a model performance indicator corresponding to the initial line loss interval determination model; The target line loss interval determination model is determined based on at least one model performance indicator.

8. A device for determining a power line loss interval determination model, characterized in that: include: A data collection module, used to collect a preset amount of power data, wherein the power data includes historical power data and current power data; A simulation graph training module, used for mapping the historical power data into a three-dimensional space to obtain a power simulation graph, and performing offline training on a pre-established convolutional neural network model based on the historical power data and the power simulation graph to obtain an initial power line loss interval determination model; The model determination module is used to perform online training on the initial power line loss interval determination model based on the current power data to obtain a target power line loss interval determination model.

9. An electronic device, characterized in that: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the method for determining the power line loss interval determination model according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the method for determining a power line loss interval determination model according to any one of claims 1 to 7 when executed.