Electric energy metering misalignment evaluation method and system considering multi-source measurement data fusion

Through comparative learning and small sample learning methods, the sample feature distribution representation learning model and complementary distance measurement function are constructed, which solves the problem of difficulty in identifying improper power metering data under the grid of new energy, and realizes efficient evaluation of improper power metering and reduces improper power metering The dependence on labeled data has been improved, and the accuracy and feasibility of the evaluation has been improved.

CN119004389AActive Publication Date: 2024-11-22STATE GRID SHANDONG ELECTRIC POWER CO MARKETING SERVICE CENT (MEASURING CENT) +1

Patent Information

Application Number
CN202411472778.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-22
Publication Date
2024-11-22
Estimated Expiration
2044-10-22

AI Technical Summary

Technical Problem

The existing technology is difficult to identify the inaccurate identification of electrical energy measurement data under the grid connection of new energy. The traditional supervised learning model requires a large amount of data labeled, which has high cost and reduced accuracy, and has limited distance measurement methods. It has failed to effectively process multi-source measurement data. Class unbalanced and metrological abnormalities.

Method used

Using the method of comparative learning combined with small sample learning, the sample feature distribution representation learning model and mean grouping category feature extractor is used, and the comparative learning and meta-learning ideas are used to construct complementary distance measurement functions, reduce dependence on labeled data, and improve regional applicability. , identify whether the power metering data is inaccurate.

Benefits of technology

Efficient use of data without explicit labels, significantly improve data utilization, improve the accuracy and feasibility of power meter misalignment evaluation, and solve the difficulty of identifying power meter misalignment and operating error calculation model caused by new energy grid connection. dynamic issues.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119004389A_ABST
    Figure CN119004389A_ABST
Patent Text Reader

Abstract

The invention provides an electric energy metering misalignment evaluation method and system considering multi-source measurement data fusion, and relates to the technical field of electric power metering evaluation, and the method comprises the steps: obtaining multi-source electric energy metering data, and carrying out the preprocessing of the multi-source electric energy metering data as an initial sample set; extracting sample feature data based on the initial sample set, and constructing a sample feature distribution representation learning model based on the sample feature data; setting a grouping similarity threshold, grouping the sample feature data, and calculating grouping category mean value sample feature data; constructing a complementary distance metric function; obtaining normal sample feature data based on the sample feature distribution representation learning model; inputting the normal sample feature data into a region-oriented mean grouping category feature extractor to obtain sample features; calculating a complementary distance metric value based on a complementary distance metric function, and inputting a softmax function to output a recognition sample category; and comparing the identified sample category with an actual user sample category, and judging whether the electric energy metering data are misaligned or not.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present disclosure relates to the technical field of electric power metering assessment, and in particular to an electric power metering inaccuracy assessment method and system considering multi-source measurement data fusion. Background Art

[0002] The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute prior art.

[0003] With the construction of new power systems, high-proportion access to new energy, and large-scale application of power electronic equipment, higher requirements are put forward for the rapid response, transmission error, accurate calibration, inaccuracy judgment, and reliable collection of measuring devices. The accuracy of power system measurement is directly related to the fairness of power market transactions. Especially in the intermittent and fluctuating output of new energy and the high-speed switching working mode of power electronic equipment, complex information such as continuous fundamental wave dynamics, wide range changes, harmonic reactive power, and wide frequency domain will be generated. The current metering device has insufficient signal response and anti-interference capabilities, resulting in the degradation of the quality of massive multi-source power measurement data, which in turn makes it difficult to carry out upper-level data fusion applications.

[0004] For example, the existing distributed photovoltaic grid-connected peak-shaving control strategy is formulated based on massive new energy grid-connected measurement data. This strategy aims at the power peak and valley problem caused by the load change of the distribution network, and proposes a segmented operation mode to ensure the stable operation of the power grid. However, the premise for the implementation of this strategy is the accuracy of the electric energy metering data, and the article does not involve the problem of inaccurate data processing of electric energy metering data. At the same time, in the abnormal data recognition model based on traditional supervised learning, the model usually requires a large amount of labeled data for training, which is not only time-consuming and labor-intensive, but also costly; and for the new energy grid connection, it also makes it difficult to identify the inaccuracy of the electric energy meter, the operation error calculation model presents dynamics, and the accuracy of inaccurate recognition decreases; the traditional distance measurement method also has its own limitations, and there are still problems such as the multi-class imbalance, metering anomalies and high cost of labeled samples that are not fully considered in multi-source measurement data. Summary of the invention

[0005] In order to solve the above problems, the present disclosure proposes an electric energy metering inaccuracy assessment method and system considering the fusion of multi-source measurement data, adopts contrastive learning combined with small sample learning to compare the similarities and differences between samples, and constructs a sample feature distribution representation learning model and a mean grouping category feature extractor based on sufficient sample data, and then performs region-oriented parameter fine-tuning to improve regional applicability. It is capable of learning the intrinsic characteristics of data without explicit labels, extracting useful information from a large amount of unlabeled data, and reducing dependence on labeled data.

[0006] According to some embodiments, the present disclosure adopts the following technical solutions: The energy metering inaccuracy assessment method considering multi-source measurement data fusion includes: Obtain multi-source electric energy metering data and use it as an initial sample set after preprocessing; Extracting sample feature data based on the initial sample set, and building a sample feature distribution representation learning model based on the sample feature data; The sample feature data is grouped based on the set grouping similarity threshold, and the mean sample feature data of the grouping category is calculated and obtained; based on the mean sample feature data of the grouping category and the sample feature data, a complementary distance measurement function is constructed; Based on a small amount of normal sample data from users in different regions, the sample features are obtained based on the sample feature distribution representation learning model; Based on the complementary distance measurement function, the complementary distance measurement value of the sample feature and the group category mean sample feature data is calculated to obtain the identified sample category; the identified sample category is compared with the actual user sample category to obtain the result of whether the electric energy metering data is inaccurate.

[0007] According to some embodiments, the present disclosure adopts the following technical solutions: The electric energy metering inaccuracy assessment system considering multi-source measurement data fusion includes: A model building module is used to obtain multi-source electric energy metering data and use it as an initial sample set after preprocessing; extract sample feature data based on the initial sample set, and build a sample feature distribution representation learning model based on the sample feature data; The category grouping module is used to group the sample feature data based on the set grouping similarity threshold, and calculate and obtain the group category mean sample feature data; A metric function construction module, used to construct a complementary distance metric function based on the grouping category mean sample feature data and the sample feature data; The evaluation module obtains sample features based on a small amount of normal sample data of users in different areas and a sample feature distribution representation learning model; calculates the complementary distance measurement value of the sample features and the grouping category mean sample feature data based on the complementary distance measurement function to obtain the identified sample category; compares the identified sample category with the actual user sample category to obtain the result of whether the electricity metering data is inaccurate.

[0008] According to some embodiments, the present disclosure adopts the following technical solutions: A computer program product comprises a computer program, wherein when the computer program is executed by a processor, the method for evaluating electric energy metering inaccuracy considering multi-source measurement data fusion is implemented.

[0009] According to some embodiments, the present disclosure adopts the following technical solutions: A non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by a processor, the electric energy metering inaccuracy assessment method considering multi-source measurement data fusion is implemented.

[0010] According to some embodiments, the present disclosure adopts the following technical solutions: An electronic device comprises: a processor, a memory and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device implements the electric energy metering inaccuracy assessment method considering multi-source measurement data fusion.

[0011] Compared with the prior art, the present invention has the following beneficial effects: The disclosed method for assessing inaccurate energy metering by considering the fusion of multi-source measurement data aims at solving the problems existing in traditional supervised learning. The disclosed method adopts contrastive learning to efficiently utilize data, especially unlabeled data. Contrastive learning can learn the intrinsic characteristics of data without explicit labels by comparing the similarities and differences between samples. It can extract useful information from a large amount of unlabeled data, reduce the dependence on labeled data, and thus significantly improve the utilization rate of data.

[0012] The disclosed method for evaluating the inaccuracy of electric energy metering considering the fusion of multi-source measurement data adopts small sample learning to evaluate the inaccuracy of electric energy metering. By borrowing the ideas of contrastive learning and meta-learning, a sample feature distribution representation learning model and a mean grouping category feature extractor are first constructed based on sufficient sample data, and then the parameters are fine-tuned for the region to improve the regional applicability. Subsequently, based on a small amount of sample data provided by the region, the samples are mapped to a feature space, and the feature centers of the samples and the space are measured. Whether the metering is inaccurate is identified according to the accuracy of sample classification. In particular, it shows high feasibility and effectiveness in processing time-varying small sample data, and solves the problems of difficulty in identifying the inaccuracy of electric energy meters caused by the grid connection of new energy, the dynamicity of the operation error calculation model, and the decrease in the accuracy of inaccuracy identification.

[0013] The electric energy metering inaccuracy assessment method considering multi-source measurement data fusion disclosed in the present invention adopts a complementary distance measurement method, and considers the influence of cosine similarity and Euclidean distance on feature measurement at the same time, measures the features simultaneously from two aspects: angle and geometric space distance, and expresses the local measurement distance of the features jointly through a measurement function as the final measurement result of the sample data, thereby realizing accurate classification of the sample data and solving the respective limitations of traditional distance measurement methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The accompanying drawings constituting a part of the present disclosure are used to provide a further understanding of the present disclosure. The illustrative embodiments of the present disclosure and their descriptions are used to explain the present disclosure and do not constitute an improper limitation on the present disclosure.

[0015] Figure 1 is an overall flow chart of the misalignment assessment method of the embodiment of the present disclosure; Figure 2 This is an architecture diagram of the comparative learning method of an embodiment of the present disclosure. DETAILED DESCRIPTION

[0016] The present disclosure is further described below in conjunction with the accompanying drawings and embodiments.

[0017] It should be noted that the following detailed descriptions are all illustrative and are intended to provide further explanation of the present disclosure. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present disclosure belongs.

[0018] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present disclosure. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0019] Example 1 In one embodiment of the present disclosure, a method for evaluating inaccurate power metering considering multi-source measurement data fusion is provided. The method takes into full consideration the problems of multi-class imbalance, measurement anomaly and high cost of labeled samples existing in multi-source measurement data, and adopts contrastive learning combined with small sample learning to construct a method for evaluating inaccurate power metering considering multi-source measurement data fusion. The method improves the quality and availability of multi-source measurement data through model parameter adaptation, feature extraction, similarity measurement, and outlier detection and elimination, and provides high-quality data support for provincial measurement data fusion applications. The method steps include: Step 1: Obtain multi-source electric energy metering data and use it as an initial sample set after preprocessing; Step 2: extracting sample feature data based on the initial sample set, and using a similarity function to respectively calculate the distance measurement values ​​of the sample feature data based on the sample feature data, and obtaining a contrastive learning loss function, thereby constructing a sample feature distribution representation learning model; Step 3: Set the group similarity threshold, group the sample feature data, and calculate the average value of the sample features as the group category mean sample feature data; Step 4: Construct a complementary distance measurement function based on the grouping category mean sample feature data and sample feature data; Step 5: Obtain a small amount of normal sample data from users in different regions, and input it into the sample feature distribution representation learning model to obtain normal sample feature data; input the normal sample feature data into the region-oriented mean grouping category feature extractor to obtain sample features; Step 6: Based on the complementary distance measurement function, calculate the complementary distance measurement value of the sample feature and the group category mean sample feature data, input the softmax function output to identify the sample category; compare the identified sample category with the actual user sample category to determine whether the electricity metering data is inaccurate.

[0020] As an embodiment, the present disclosure uses the electricity information collection system and the marketing business application system as the data processing platform, and uses the distributed photovoltaic power metering collection data as a modeling implementation business example. The specific implementation process is as follows: Step 1: Based on the new generation of electricity consumption information collection system, multi-source electricity metering data is obtained. The multi-source electricity metering data is multi-source distributed photovoltaic electricity metering data. After being summarized and sorted, it is pre-processed. During the pre-processing, data cleaning is first performed, specifically including deleting duplicate data and missing data, and then data integration is performed to merge multiple data sources into one data storage, delete redundant data, and then perform data normalization to eliminate dimensional differences and complete the construction of the initial sample set; the multi-source distributed photovoltaic electricity metering data categories include: distributed photovoltaic operation mode, photovoltaic business user type, electricity metering and environmental factors. The details are shown in Table 1 below.

[0021] Table 1 Distributed photovoltaic multi-source electricity metering data categories

[0022] Step 2: extracting sample feature data based on the initial sample set, and using a similarity function to calculate the distance metric values ​​of the sample feature data based on the sample feature data, and obtaining a contrastive learning loss function, thereby constructing a sample feature distribution representation learning model; Among them, based on the initial sample set, the encoder is combined with the attention mechanism to extract sample feature data, and the sample feature distribution representation learning model initially constructed based on contrastive learning is input, so that the feature distance of samples of the same category is shortened and the feature distance of samples of different categories is extended, and the spatial distribution of the feature data of the electric energy metering samples is obtained. The specific process is as follows: Step 21: Based on the initial sample set, the encoder is used in combination with the attention mechanism to extract sample feature data; The distributed photovoltaic operation mode, photovoltaic business user type, and environmental factors of the initial sample set are used as the input sample data of the attention mechanism, and the electric energy metering data of the initial sample set is used as the input sample data of the encoder. After the encoder is combined with the attention mechanism, the encoded data is output, and the encoded data is input into the generator. Combined with the distributed photovoltaic operation mode, photovoltaic business user type, and attention mechanism influencing parameters of the same initial sample set, the reproduced sample data is output as the sample feature data. Feedback and parameter adjustment are performed through the loss function until the error between the input sample data and the reproduced sample data is less than the set threshold. After the training is completed, the encoder combined with the attention mechanism is output as the sample feature data extraction model.

[0023] Step 22: Based on the initial sample set, the sample feature data is input into the sample feature distribution representation learning model initially constructed based on contrastive learning, and the feature distance measurement values ​​are calculated respectively using the similarity function to obtain the contrastive learning loss function; Among them, the sample feature distribution representation learning model structure constructed based on contrastive learning is as follows: Figure 2 As shown in the figure, it includes two data processing branches, the upper and lower branches, both of which include an encoder combined with an attention mechanism (Attention-Encoder) module, a feature data module, a similarity function calculation module, and a loss function calculation module. The upper and lower branches are symmetrical and can share parameters.

[0024] Among them, the encoder is combined with the attention mechanism (Attention-Encoder) module to construct the feature data extraction model for step 21.

[0025] Among them, the sample data are input into the sample feature distribution representation learning model initially constructed based on contrastive learning. The specific data processing process includes two categories. The sample data of the same user at different times are respectively input into the upper and lower branches to obtain the sample feature data. Through the similarity function calculation and loss function feedback, the feature data extracted from the sample set of the same user are shortened in the projection space; the sample data of different users are respectively input into the upper and lower branches to obtain the sample feature data. Through the similarity function calculation and loss function feedback, the feature data extracted from the sample sets of different users are pushed further in the projection space.

[0026] Specifically, two examples of electricity metering sample data for the same user are represented as and , the two different user energy metering sample data are represented as and , the sample feature data extraction model is used to encode the energy metering sample data into a vector:

[0027]

[0028]

[0029] in, It represents the sample feature data extraction model constructed by combining the encoder with the attention mechanism. , , is the encoded sample feature vector.

[0030] Calculate energy metering value using InfoNCE loss Eigenvector and electric energy measurement value Contrastive learning loss function between feature vectors :

[0031] Among them, the dot multiplication method is used to and Calculate the similarity between represents the temperature hyperparameter, which is the sample difficulty recognition penalty coefficient, and increases the distance between positive samples and negative samples with high similarity; N represents the number of different types of sample groups input; For the encoded Feature vectors of samples of different classes; Step 23: Update the encoder parameters and attention mechanism parameters according to the back propagation of the contrastive learning loss function until the loss function is less than the set threshold, then fix the model parameters and complete the construction of the sample feature distribution representation learning model.

[0032] Step 3: Set the group similarity threshold, group the sample feature data, and calculate the average value of the sample features as the group category mean sample feature data; In view of the fact that the characteristic distribution distance of similar distributed photovoltaic users in the same area is often small, a grouping similarity threshold B is set, and users whose distance is smaller than the threshold are grouped into the same group. K groups, and calculate the feature averages as k The mean sample characteristic data of each grouping category; The mean sample characteristics of the grouped categories are :

[0033] in, Represents the sample feature vector matrix, containing K features; L is the total number of samples in the group, Indicatesl Sample K The value vector of the features; .

[0034] Step 4: Construct a complementary distance measurement function based on the grouped category mean sample feature data and sample feature data; Furthermore, based on the sample feature data of the grouped category mean and the sample feature data of electric energy metering, the cosine distance and Euclidean distance are used to construct a complementary distance measurement function to evaluate the similarity of small sample learning classification. The specific process is as follows: Step 41: Based on the grouping category mean sample feature data and the to-be-classified electric energy metering sample feature data, the cosine similarity algorithm is used to calculate the measurement coefficient ;

[0035] in, represents the sample feature vector; represents the electric energy metering sample to be classified; k represents the mean sample of the kth group; The sample feature vector representing the electric energy metering sample q to be classified; Represents the sample feature vector of the mean sample k of the kth group; Represents the cosine similarity calculation function, obtain and The cosine similarity value of ; Represents the threshold activation function, whose value range is (0,1).

[0036] Step 42: Use the Euclidean distance function combined with the metric coefficient to construct a complementary distance metric function ;

[0037] in, Used for balance and The Euclidean distance size is: Step 43: Based on the grouping category mean sample feature data and the sample feature data of electric energy metering, the complementary distance metric value is calculated using the complementary distance metric function, input into the softmax function, and the grouping category of the electric energy metering sample is output. y .

[0038]

[0039] in, b is a training parameter, initialized to 0; softmax is a multi-classification probability function, the probability range of each vector is (0,1), and the sum of the probability values ​​of each vector is 1; the group category with the highest probability value in the softmax vector isy Category of; If the sample measurement data is normal, it will be grouped into the correct grouping category of the user; if the sample measurement data is abnormal, it will be grouped into other groups. Incorrect grouping categories indicate abnormal measurement data.

[0040] Step 5: Obtain a small amount of normal sample data from users in different regions, and input it into the sample feature distribution representation learning model to obtain normal sample feature data; input the normal sample feature data into the region-oriented mean grouping category feature extractor to obtain sample features; Among them, the region-oriented mean grouping category feature extractor uses a complementary distance measurement function to calculate the distance between a small amount of normal sample data of users in different regions and the mean grouping category sample features, and feeds it back to the initial sample feature distribution representation learning model constructed based on contrastive learning constructed in step 22. By feeding back fine-tuning parameters, the fine-tuning parameters include the encoder combined with the attention mechanism (Attention-Encoder) module, the similarity function calculation module, the loss function calculation module, and the parameters of the softmax function in step 43, so as to achieve the applicability of the model to the region.

[0041] Step 6: Based on the complementary distance measurement function, the sample features and the grouping category mean sample feature data are input into the complementary distance measurement function, the complementary distance measurement value of the sample features and the grouping category mean sample feature data is calculated, and the softmax function is input to output the identified sample category; the identified sample category is compared with the actual user sample category to determine whether the electric energy metering data is inaccurate.

[0042] If the identified sample category is the same as the actual user sample category, the electric energy metering data is normal and the normal electric energy metering data is used for subsequent load forecasting tasks; if the identified sample category is different from the actual user sample category, the electric energy metering data is abnormal and the abnormal data is eliminated.

[0043] Simulation experiment In order to verify the effectiveness of the method proposed in this disclosure for evaluating the misalignment of electric energy metering, a comparative description is given of the existing research methods for identifying misalignment of abnormal photovoltaic metering. The comparative methods are as follows: (1) k-means clustering comparison method: PV users in the same area are selected and the PV power metering inaccuracy is evaluated based on the sample distance deviation value; (2) SVM: A photovoltaic power metering inaccuracy assessment model built using support vector machine (SVM); (3) LSTM: A photovoltaic power metering inaccuracy assessment model built using the Long Short-Term Memory (LSTM) algorithm; (4) Methods of embodiments of the present disclosure; In order to demonstrate the generalization applicability of the method disclosed in the present invention, photovoltaic power metering user data in three different regions are collected to construct three groups of data sample sets as follows: Data set 1: contains 861 samples from region A; Data set 2: 59 samples from area B were obtained; Data set 3: 902 samples from area C were obtained; The accuracy verification results of the method proposed in this disclosure are shown in Table 2.

[0044] Table 2 Comparison of accuracy of electric energy metering inaccuracy assessment

[0045] As can be seen from Table 2, in simulation data groups 1 and 3, the disclosed method shows superior recognition performance, can effectively track the dynamic changes in photovoltaic power generation, and identify abnormal data in electric energy metering. Especially in data group 2, in a small sample data set, it is still possible to adjust parameters based on the pre-trained model, thereby accurately identifying abnormal metering data.

[0046] As an embodiment, the electric energy metering inaccuracy assessment method considering multi-source measurement data fusion disclosed in the present invention is implemented based on an electric energy metering inaccuracy data recognition model. The training process of the electric energy metering inaccuracy data recognition model is as follows: Figure 1 As shown, specifically: 1) Use the electricity consumption information collection system to obtain multi-source measurement data and construct an initial sample set after preprocessing; 2) Based on the initial sample set, an encoder combined with an attention mechanism is used to extract sample feature data, which is then input into a sample feature distribution representation learning model based on contrastive learning, thereby shortening the feature distances of samples of the same category and increasing the feature distances of samples of different categories, thereby obtaining the spatial distribution of the features of the energy metering samples; 3) Set the grouping similarity threshold and group them, treat the sample feature data that is less than the grouping similarity threshold as the same grouping category, and calculate the feature average as the grouping category mean sample feature data; 4) Based on the sample feature data of the mean value of the grouped categories and the sample feature data of the electric energy metering, the cosine distance and the Euclidean distance are used to construct a complementary distance measurement function to evaluate the similarity of small sample learning classification; 5) A small amount of normal sample data of users in different regions is used as a support sample set, and is input into a sample feature distribution representation learning model built based on contrastive learning, and the data exceeding the threshold is output; 6) Based on the data exceeding the threshold, the complementary distance metric function is used to calculate the distance from the sample features of the mean grouping category, and the fine-tuning encoder, contrastive learning and softmax function are fed back for parameter fine-tuning, and the model is updated to output the data category exceeding the threshold again; 7) Finally, the trained model for identifying inaccurate electric energy metering data is output, and the electric energy metering feature data to be evaluated is obtained for identification and evaluation.

[0047] Example 2 In one embodiment of the present disclosure, a system for evaluating electric energy metering inaccuracy considering multi-source measurement data fusion is provided, comprising: A model building module is used to obtain multi-source electric energy metering data and use it as an initial sample set after preprocessing; extract sample feature data based on the initial sample set, and build a sample feature distribution representation learning model based on the sample feature data; The category grouping module is used to group the sample feature data based on the set grouping similarity threshold, and calculate and obtain the group category mean sample feature data; A metric function construction module, used to construct a complementary distance metric function based on the grouping category mean sample feature data and the sample feature data; The evaluation module obtains sample features based on a small amount of normal sample data of users in different areas and a sample feature distribution representation learning model; calculates the complementary distance measurement value of the sample features and the grouping category mean sample feature data based on the complementary distance measurement function to obtain the identified sample category; compares the identified sample category with the actual user sample category to obtain the result of whether the electricity metering data is inaccurate.

[0048] Example 3 In one embodiment of the present disclosure, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the method for evaluating electric energy metering inaccuracy considering multi-source measurement data fusion is implemented.

[0049] Example 4 In one embodiment of the present disclosure, a non-transitory computer-readable storage medium is provided, wherein the non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by a processor, the electric energy metering inaccuracy assessment method considering multi-source measurement data fusion is implemented.

[0050] Example 5 An electronic device comprises: a processor, a memory and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device implements the electric energy metering inaccuracy assessment method considering multi-source measurement data fusion.

[0051] The present disclosure 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 disclosure. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks 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 generate 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.

[0052] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0053] Although the above describes the specific implementation methods of the present disclosure in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present disclosure. Technical personnel in the relevant field should understand that on the basis of the technical solution of the present disclosure, various modifications or variations that can be made by those skilled in the art without creative work are still within the scope of protection of the present disclosure.

Claims

1. An electric energy metering inaccuracy assessment method considering multi-source measurement data fusion is characterized in that: include: Obtain multi-source electric energy metering data and use it as an initial sample set after preprocessing; Extracting sample feature data based on the initial sample set, and building a sample feature distribution representation learning model based on the sample feature data; The sample feature data is grouped based on the set grouping similarity threshold, and the mean sample feature data of the grouping category is calculated and obtained; based on the mean sample feature data of the grouping category and the sample feature data, a complementary distance measurement function is constructed; Based on a small amount of normal sample data from users in different regions, the sample features are obtained based on the sample feature distribution representation learning model; Based on the complementary distance measurement function, the complementary distance measurement value of the sample feature and the group category mean sample feature data is calculated to obtain the identified sample category; the identified sample category is compared with the actual user sample category to obtain the result of whether the electric energy metering data is inaccurate.

2. The method for evaluating electric energy metering inaccuracy considering multi-source measurement data fusion according to claim 1, characterized in that: The multi-source electric energy metering data obtained is multi-source distributed photovoltaic electric energy metering data, and the multi-source distributed photovoltaic electric energy metering data categories include: distributed photovoltaic operation mode, photovoltaic business user type, electric energy metering and environmental factors.

3. The method for evaluating electric energy metering inaccuracy considering multi-source measurement data fusion according to claim 1, characterized in that: The extracting sample feature data based on the initial sample set includes: The encoder is used in combination with the attention mechanism to extract sample feature data. The distributed photovoltaic operation mode, photovoltaic business user type, and environmental factors of the initial sample set are used as the input sample data of the attention mechanism. The electric energy metering data of the initial sample set is used as the input sample data of the encoder, and the encoded data is output. The encoded data is input into the generator, and the distributed photovoltaic operation mode, photovoltaic business user type, and attention mechanism influencing parameters of environmental factors output by the attention mechanism are combined to output the reproduced sample data as the sample feature data.

4. The method for evaluating electric energy metering inaccuracy considering multi-source measurement data fusion according to claim 1, characterized in that: Based on the sample feature data, the similarity function is used to calculate the distance measurement value of the sample feature data respectively, and the contrast learning loss function is obtained, so as to construct the sample feature distribution representation learning model, including: The sample feature data is input into the initial sample feature distribution representation learning model constructed based on contrastive learning. The feature distance measurement values ​​are calculated using the similarity function to obtain the contrastive learning loss function. The encoder parameters and attention mechanism parameters are updated by backpropagation according to the contrastive learning loss function until the loss function is less than the set threshold. The model parameters are fixed to complete the construction of the final sample feature distribution representation learning model.

5. The method for evaluating electric energy metering inaccuracy considering multi-source measurement data fusion according to claim 1, characterized in that: Based on the grouping category mean sample feature data and the sample feature data, a complementary distance measurement function is constructed, including: based on the grouping category mean sample feature data and the sample feature data, a measurement coefficient is calculated using cosine similarity, in, represents the sample feature vector; Indicates the electric energy metering samples to be classified; k Indicates k Class grouping mean sample; The sample feature vector representing the electric energy metering sample q to be classified; Represents the mean sample of the k-th group k The sample feature vector of Represents the cosine similarity calculation function, obtain and The cosine similarity value of ; Represents the threshold activation function, whose value range is (0,1).

6. The method for evaluating electric energy metering inaccuracy considering multi-source measurement data fusion according to claim 5, characterized in that: The Euclidean distance function is combined with the metric coefficient to construct a complementary distance metric function, including: in, Used for balance and The Euclidean distance size.

7. The electric energy metering inaccuracy assessment system considering multi-source measurement data fusion is characterized by: include: The model building module is used to obtain multi-source electric energy metering data and use it as an initial sample set after preprocessing; Extracting sample feature data based on the initial sample set, and building a sample feature distribution representation learning model based on the sample feature data; The category grouping module is used to group the sample feature data based on the set grouping similarity threshold, and calculate and obtain the group category mean sample feature data; A metric function construction module, used to construct a complementary distance metric function based on the grouping category mean sample feature data and the sample feature data; The evaluation module obtains sample features based on a small amount of normal sample data of users in different areas and a sample feature distribution representation learning model; calculates the complementary distance measurement value of the sample features and the grouping category mean sample feature data based on the complementary distance measurement function to obtain the identified sample category; compares the identified sample category with the actual user sample category to obtain the result of whether the electricity metering data is inaccurate.

8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for evaluating electric energy metering inaccuracy considering multi-source measurement data fusion as described in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by the processor, the method for evaluating electric energy metering inaccuracy considering multi-source measurement data fusion as described in any one of claims 1 to 6 is implemented.

10. An electronic device, characterized in that: include: A processor, a memory and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory so that the electronic device executes the electric energy metering inaccuracy assessment method considering multi-source measurement data fusion as described in any one of claims 1-6.

Citation Information

Patent Citations

  • Electric energy metering device state assessment method, device and system

    CN106707224A

  • Power demand prediction device and method

    JP2017146815A

Cited By

  • Pig growth trait prediction method based on SNP grouping selection and ensemble learning

    CN121662150A