CN111461923A - Electricity stealing monitoring system and method based on deep convolutional neural network - Google Patents

Electricity stealing monitoring system and method based on deep convolutional neural network Download PDF

Info

Publication number
CN111461923A
CN111461923A CN202010259498.9A CN202010259498A CN111461923A CN 111461923 A CN111461923 A CN 111461923A CN 202010259498 A CN202010259498 A CN 202010259498A CN 111461923 A CN111461923 A CN 111461923A
Authority
CN
China
Prior art keywords
neural network
data
convolutional neural
deep convolutional
monitoring
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202010259498.9A
Other languages
Chinese (zh)
Inventor
吴健
孙伟
吴奎华
张宇帆
杨波
冯亮
崔灿
杨杨
刘蕊
艾芊
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shanghai Jiaotong University
State Grid Corp of China SGCC
Liaocheng Power Supply Co of State Grid Shandong Electric Power Co Ltd
Original Assignee
Shanghai Jiaotong University
State Grid Corp of China SGCC
Liaocheng Power Supply Co of State Grid Shandong Electric Power Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shanghai Jiaotong University, State Grid Corp of China SGCC, Liaocheng Power Supply Co of State Grid Shandong Electric Power Co Ltd filed Critical Shanghai Jiaotong University
Priority to CN202010259498.9A priority Critical patent/CN111461923A/en
Publication of CN111461923A publication Critical patent/CN111461923A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Electricity, gas or water supply
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"

Abstract

The invention discloses an electricity stealing monitoring system and method based on a deep convolutional neural network, which relate to the technical field of power monitoring and comprise a data input module, an electricity stealing deep convolutional neural network model training module based on the deep convolutional neural network and an electricity stealing deep convolutional neural network model monitoring effect evaluation module, wherein the data input module synthesizes normal electricity utilization data and electricity stealing data, preprocesses the synthesized data, and divides the preprocessed data into a training set and a test set; the electric larceny depth convolution neural network model training module is used for receiving data preprocessed by the data input module and training the electric larceny depth convolution neural network model; the invention realizes the mining of effective information contained in data, improves the electricity stealing monitoring accuracy and has robustness for normal electricity load change.

Description

Electricity stealing monitoring system and method based on deep convolutional neural network
Technical Field
The invention relates to the technical field of power monitoring, in particular to a system and a method for monitoring electricity stealing based on a deep convolutional neural network.
Background
Thanks to the latest information technology, recent years witness to the emerging and rapid development of industrial internet of things for intelligent energy. The intelligent instrument is used as a terminal device in the Internet of things and has the responsibility of accurately recording the consumption of electric energy. However, the use of digital smart meters introduces new possibilities for energy theft. At present, solutions to the problem of electricity stealing can be divided into three categories, namely, methods based on state estimation, game theory and data mining. The data mining-based detection method utilizes machine learning techniques to extract statistical patterns of energy consumption data. With the benefit of recent development of machine learning techniques, such an approach is expected to achieve high-performance electricity theft monitoring.
The document Jokar P, Arianpoo N, L eung V C M.electric the detection in the amplitude vector computers' control patterns [ J ]. IEEE Transactions on Smart Grid,2016,7(1):216-226 reports a power stealing monitoring method based on a support vector machine (SVC) which has high classification performance and strong robustness to the change of a normal power mode.
Therefore, those skilled in the art are devoted to develop a system and method for monitoring electricity stealing based on a deep convolutional neural network to seek a large data monitoring method which can effectively cope with the needs of electricity stealing monitoring due to the increasing data measurement frequency and the increasing data scale brought by the wide deployment of smart meters.
Disclosure of Invention
In view of the above defects in the prior art, the technical problems to be solved by the present invention are the problem that indexes such as monitoring accuracy and calculation efficiency are low in a shallow machine learning electricity stealing monitoring model represented by an SVM, and the problem that the shallow machine learning electricity stealing monitoring model depends on a CPU to calculate the existing calculation time consumption and cannot meet the requirement of electricity stealing monitoring real-time performance.
In order to achieve the purpose, the invention provides an electricity stealing monitoring system based on a deep convolutional neural network, which comprises a data input module, an electricity stealing deep convolutional neural network model training module based on the deep convolutional neural network, and an electricity stealing deep convolutional neural network model monitoring effect evaluation module, wherein the data input module synthesizes normal electricity utilization data and electricity stealing data, preprocesses the synthesized data, and divides the preprocessed data into a training set and a test set; the electric larceny depth convolution neural network model training module is used for receiving data preprocessed by the data input module and training the electric larceny depth convolution neural network model; and the electricity stealing depth convolution neural network model monitoring effect evaluation module is used for comprehensively evaluating the electricity stealing monitoring effect of the electricity stealing depth convolution neural network model.
Further, the data preprocessing method of the data input module is that firstly, the synthesized data set is a matrix X of m × n,
Figure BDA0002438759120000021
where m is the total number of training samples, n is the number of input features, xijA jth feature representing an ith sample of input data;
then, carrying out normalization processing on the data of each column of the original data set matrix:
Figure BDA0002438759120000022
wherein the content of the first and second substances,
Figure BDA0002438759120000023
for the minimum value of each column of data,
Figure BDA0002438759120000024
is the level difference of each column of data.
Furthermore, the electricity stealing deep convolutional neural network model based on the deep convolutional neural network comprises an electricity stealing deep convolutional neural network model classifier which is composed of a series of convolutional layers, pooling layers and full-link layers.
Further, the electricity stealing deep convolution neural network model classifier distinguishes malicious electricity utilization behaviors from normal electricity utilization behaviors by extracting input features of a user electricity utilization curve.
Furthermore, the convolution layer is composed of convolution kernels with a learning parameter function, input features are effectively extracted, in the forward propagation process, the convolution kernels use input quantities to carry out convolution calculation, and the convolution kernels move on the result of data preprocessing on the historical monitoring data according to a predefined step length, namely the distance of each movement of the convolution kernels, so that local features of the historical monitoring data are reserved and more abstract features are extracted.
Further, the pooling layer reduces the dimensionality of the features while retaining important information.
Furthermore, the full connection layer is composed of rearranged neurons and is positioned at the tail end of the power stealing deep convolution neural network model architecture.
Further, the monitoring effect evaluation module of the electricity stealing deep convolution neural network model evaluates the accuracy, recall rate, specificity, F1 score and accuracy index of electricity stealing monitoring.
In order to achieve the above object, the present invention further provides a method for monitoring electricity stealing based on a deep convolutional neural network, comprising the following steps:
s1, acquiring monitoring data;
s2, preprocessing the acquired data;
s3, inputting the preprocessed data into a power stealing detection deep convolution neural network model;
and S4, outputting a judgment result, and if the judgment result is consistent with the electricity stealing model data, giving an alarm to the data.
Further, inputting the preprocessed data into the deep convolutional neural network model for detecting electricity stealing includes the following steps:
s3-1, acquiring historical monitoring data;
s3-2, carrying out data preprocessing on the historical monitoring data;
s3-3, training a power stealing monitoring deep convolution neural network model until convergence;
and S3-4, evaluating the electricity stealing monitoring deep convolutional neural network model according to the confusion matrix evaluation index.
Compared with the prior art, the electricity stealing monitoring method based on the deep convolutional neural network is a data-driven electricity stealing monitoring method compared with the traditional electricity stealing monitoring method, and the effective information contained in the data is mined by utilizing the stronger nonlinear mapping and feature extraction capability of a deep learning network; compared with a typical electricity stealing monitoring algorithm SVM, the method provided by the invention has the advantages that the electricity stealing monitoring accuracy is improved, and the method is more robust to normal electricity load change.
The conception, the specific structure and the technical effects of the present invention will be further described with reference to the accompanying drawings to fully understand the objects, the features and the effects of the present invention.
Drawings
FIG. 1 is a model architecture diagram of a deep convolutional neural network-based electricity stealing monitoring system in accordance with a preferred embodiment of the present invention;
FIG. 2 is typical electricity stealing data for a preferred embodiment of the invention;
FIG. 3 is a method for monitoring electricity stealing based on deep convolutional neural network according to a preferred embodiment of the present invention;
FIG. 4 is a training and testing process of a deep convolutional neural network model for power stealing in accordance with a preferred embodiment of the present invention;
FIG. 5 is a diagram illustrating the dynamic variation of the training results of the deep convolutional neural network model in accordance with a preferred embodiment of the present invention;
FIG. 6 is a confusion matrix heatmap on a test set in accordance with a preferred embodiment of the present invention.
Detailed Description
The technical contents of the preferred embodiments of the present invention will be more clearly and easily understood by referring to the drawings attached to the specification. The present invention may be embodied in many different forms of embodiments and the scope of the invention is not limited to the embodiments set forth herein.
In the drawings, structurally identical elements are represented by like reference numerals, and structurally or functionally similar elements are represented by like reference numerals throughout the several views. The size and thickness of each component shown in the drawings are arbitrarily illustrated, and the present invention is not limited to the size and thickness of each component. The thickness of the components may be exaggerated where appropriate in the figures to improve clarity.
Fig. 1 is a diagram of a model architecture of a deep convolutional neural network-based power stealing monitoring system according to a preferred embodiment of the present invention, which includes a data input module, a deep convolutional neural network model training module based on a deep convolutional neural network, and a power stealing deep convolutional neural network model monitoring effect evaluation module. The data input module is mainly used for synthesizing normal power utilization data and electricity stealing data according to the predicted electricity stealing data type, then preprocessing the synthesized data, and dividing the preprocessed data into a training set and a test set; in the electric larceny deep convolution neural network model training module, training an electric larceny deep convolution neural network model based on a Tensorflow deep learning platform; and in the electricity stealing deep convolution neural network model monitoring effect evaluation module, evaluation is carried out according to indexes such as Precision (PRE), Recall (REC), Specificity (SPE), F1 score, accuracy (Acc) and the like, so that the electricity stealing monitoring effect is comprehensively evaluated.
In actual conditions, normal data are easy to obtain, electricity stealing data are few, but electricity stealing types can be predicted, and therefore electricity stealing data can be effectively simulated. Therefore, the embodiment of the invention selects the GEFCom2012 competition provision data as a normal power consumption sample set, adopts the power consumption data of the area 1 per hour from 2004 to 2005 as normal data, and simulates the following 6 power stealing types. For a 24-hour normal electricity sample x ═ { x ═1,x2,...,x24The possible patterns of electricity stealing are as follows, t1, 2.
One, h1(x)＝αx,α＝random(0.1,0.8)
II,
Figure BDA0002438759120000041
III, h3(x)＝γ·x,γt＝random(0.1,0.8)
Fourthly, h4(x)＝γ·mean(x),γt＝random(0.1,0.8)
Fifth, h5(x)＝mean(x)
Sixth, h6(xt)＝x24-t
Where α is the random number sampled from between the intervals (0.1,0.8), β is a vector consisting of 0,1, γ is a random vector where each element is the random number sampled from between the intervals (0.1,0.8), and mean () represents the averaging operation.
Wherein, the first electricity stealing type multiplies all load curves by the same random number less than 1; the second electricity stealing type is that the reading of the intelligent ammeter is reduced to 0 within a certain specific time interval; the third type of electricity stealing is that the electricity load is multiplied by different random numbers at each moment; the fourth and fifth electricity stealing types change the load curve of a day into a straight line related to the average value; the sixth electricity stealing type is a mirror image of a load curve, and the electricity stealing mode mainly aims at different electricity prices in one day.
The methods of use of the data entry module, training set and test set on the model followed the conventional methods of neural networks proposed in the literature L i M, Zhang T, Chen Y, et al, efficient mini-batch training for storage optimization [ C ]// Proceedings of the 20th ACM SIGKDD international reference on knowledge discovery and data mining 2014:661 suite 670.
The data preprocessing method, the synthesized data set is a matrix X of m × n,
Figure BDA0002438759120000042
where m is the total number of training samples, n is the number of input features, xijRepresenting the jth feature of the ith sample input data.
Normalizing the data of each column of the original data set matrix:
Figure BDA0002438759120000051
wherein the content of the first and second substances,
Figure BDA0002438759120000052
for the minimum value of each column of data,
Figure BDA0002438759120000053
is the level difference of each column of data.
The proposed deep convolutional neural network model structure of the deep convolutional neural network consists of a series of FeatureMaps, i.e., extracted feature convolutional layers, and Pooling, i.e., pooled layers, and finally, through a full connected layer, i.e., fused connected L eyes, the decision is made by the deep convolutional neural network model classifier.
The goal of the loss function, electricity stealing depth convolution neural network model based classifier, is to output a probability distribution that matches the target distribution, which is a discrete distribution. Thus, cross entropy is used as a loss function:
H(p,q)＝-Ep[log q]
＝H(p)+DKL(p||q)
wherein H (p) is a constant, DKLThus, minimizing cross-entropy is equivalent to minimizing the Kullback-L eibler divergence, which is a measure of the difference between one probability distribution and another.
Thus, once the loss function remains stable on the training set, the power stealing deep convolutional neural network model can make a determination of power stealing.
The monitoring effect evaluation module of the electricity stealing deep convolution neural network model evaluates indexes such as accuracy (PRE), Recall (REC), Specificity (SPE), F1 score and accuracy (Acc) of electricity stealing monitoring, and all data are normalized to the range of [ -1,1 ]. The specific definition of the evaluation index according to the confusion matrix is as follows:
confusion matrix
Actual electricity stealing Is actually normal
Predicting electricity stealing TP FN
Predicting normality FP TN
Wherein, TP is the number of the actual electricity stealing data judged as electricity stealing, FN is the number of the actual electricity stealing data judged as normal, FP is the number of the actual normal data judged as electricity stealing, and TN is the number of the actual normal data judged as normal.
1) Accuracy (PRE): the accuracy is defined as the proportion of correctly predicted electricity stealing cases in all actual electricity stealing cases.
precision＝TP/(TP+FP)
2) Recall (REC): the recall is defined as the proportion of correctly predicted electricity stealing cases among all predicted electricity stealing cases.
recall＝TP/(TP+FN)
3) Specificity (SPE): specificity is defined as the proportion of correctly predicted normal use patterns in all predicted normal use patterns.
specificity＝TN/(TN+FP)
4) F1 score: the F1 score conveys a balance between accuracy and recall, with an optimal value of 1.
F1＝2×PRE×REC/(PRE+REC)
5) Accuracy (ACC): accuracy is defined as the proportion of correct classification.
accuracy＝(TP+TN)/(TP+FN+FP+TN)
Considering that the historical monitoring data has the autocorrelation characteristic, the historical monitoring data of 6 continuous days (144 hours) is used as the input of the electricity stealing deep convolution neural network model and is adjusted to be a matrix of 12 x 12. The power stealing depth convolutional neural network model includes two convolutional layers and two pooling layers to down-sample historical monitoring data. The specific architecture of the deep convolutional neural network model for power stealing is shown in table 1, and the software architecture according to the invention is a tensrflow deep learning framework.
TABLE 1 deep convolutional neural network model architecture for power stealing
Figure BDA0002438759120000061
The power stealing deep convolutional neural network model was trained using an AdampProp optimizer with a learning rate of 1e-3 and a sub-training set size of 128. for both convolutional and fully-connected layers, the RE L U nonlinear function activation was used.
As shown in fig. 2, which is typical electricity stealing data of a preferred embodiment of the present invention, Normal is Normal power consumption, and the whole day power consumption of electricity stealing type one Theft1, three Theft3 and four Theft4 is always lower than the expected energy consumption. Electricity stealing type two Theft2 remains normal for most of the day, but suddenly drops to zero when the price of electricity is high. Electricity stealing type six Theft6 reverses the order of smart meter measurements in an attempt to accommodate changes in electricity prices, i.e., less electricity usage when electricity prices are high.
As shown in fig. 3, the method for monitoring electricity stealing based on deep convolutional neural network according to a preferred embodiment of the present invention includes the following steps:
s1, acquiring monitoring data;
s2, preprocessing the acquired data;
s3, inputting the preprocessed data into a power stealing detection deep convolution neural network model;
and S4, outputting a judgment result, and if the judgment result is consistent with the electricity stealing model data, giving an alarm to the data.
As shown in fig. 4, the training and testing process of the deep convolutional neural network model for power stealing according to a preferred embodiment of the present invention includes the following steps:
s3-1, acquiring historical monitoring data;
s3-2, carrying out data preprocessing on the historical monitoring data;
s3-3, training a power stealing monitoring deep convolution neural network model until convergence;
and S3-4, evaluating the electricity stealing monitoring deep convolutional neural network model according to the confusion matrix evaluation index.
FIG. 5 is a diagram showing the dynamic variation of the training result of the deep convolutional neural network model according to a preferred embodiment of the present invention, which shows the dynamic variation of the objective function and accuracy of the deep convolutional neural network model with training times. It can be seen that the power stealing deep convolutional neural network model converges rapidly, remaining almost stable over 2000 training iterations. Therefore, the result proves that the power stealing deep convolution neural network model has good convergence performance.
The detection result of the electricity stealing monitoring method based on the deep convolutional neural network is compared with the detection result of the SVC model, as shown in table 2, it should be noted that the time in table 2 includes training time and testing time. It can be seen that the electricity stealing monitoring method based on the deep convolutional neural network can achieve higher precision on both a training set and a test set. Therefore, the proposed electricity stealing monitoring method based on the deep convolutional neural network does not have the problem of overfitting. Furthermore, the relative improvement in terms of accuracy of the test set is about 20.1% compared to the SVC method. Furthermore, with GPU acceleration, the time to train and test detectors based on the power stealing deep convolutional neural network model is much shorter than with SVC-based methods. Therefore, the electricity stealing monitoring method based on the deep convolutional neural network is suitable for real-time detection.
TABLE 2 accuracy and time of monitoring for electricity stealing
Figure BDA0002438759120000071
As shown in fig. 6, which is a confusion matrix heatmap on a test set according to a preferred embodiment of the present invention, the results of evaluating the metrics are shown in table 3, and the SVC-based method tends to classify a large number of normal behaviors as electricity stealing behaviors, so the SVC-based method is less robust to changes in the normal electricity using behaviors. Furthermore, almost all metrics indicate that the detector performance of the proposed electricity stealing deep convolutional neural network model is superior to the SVC-based detector.
TABLE 3 ELECTRICITY-STEALING MONITORING INDICATOR COMPARISON
PRE REC SPE F1
SVC method 100％ 74.75％ 100％ 0.8555
The method of the invention 99.83％ 99.92％ 99.83％ 0.9988
The foregoing detailed description of the preferred embodiments of the invention has been presented. It should be understood that numerous modifications and variations could be devised by those skilled in the art in light of the present teachings without departing from the inventive concepts. Therefore, the technical solutions available to those skilled in the art through logic analysis, reasoning and limited experiments based on the prior art according to the concept of the present invention should be within the scope of protection defined by the claims.

Claims (10)

1. A power stealing monitoring system based on a deep convolutional neural network is characterized by comprising a data input module, a power stealing deep convolutional neural network model training module based on the deep convolutional neural network and a power stealing deep convolutional neural network model monitoring effect evaluation module, wherein the data input module synthesizes normal power utilization data and power stealing data, preprocesses the synthesized data, and divides the preprocessed data into a training set and a test set; the power stealing depth convolution neural network model training module receives the data preprocessed by the data input module and trains the power stealing depth convolution neural network model; and the electricity stealing deep convolution neural network model monitoring effect evaluation module is used for comprehensively evaluating the electricity stealing monitoring effect of the electricity stealing deep convolution neural network model.
2. The deep convolutional neural network-based electricity stealing monitoring system of claim 1, wherein the data preprocessing method of the data input module first, the synthesized data set is a matrix X of m × n,
Figure FDA0002438759110000011
where m is the total number of training samples, n is the number of input features, xijA jth feature representing an ith sample of input data;
then, carrying out normalization processing on the data of each column of the original data set matrix:
Figure FDA0002438759110000012
wherein the content of the first and second substances,
Figure FDA0002438759110000013
Figure FDA0002438759110000014
for the minimum value of each column of data,
Figure FDA0002438759110000015
is the level difference of each column of data.
3. The deep convolutional neural network-based power stealing monitoring system of claim 1, wherein the deep convolutional neural network-based power stealing deep convolutional neural network model comprises a power stealing deep convolutional neural network model classifier consisting of a series of convolutional layers, pooling layers, and fully-connected layers.
4. The deep convolutional neural network-based power stealing monitoring system of claim 3, wherein the power stealing deep convolutional neural network model classifier distinguishes malicious power consumption behavior from normal power consumption behavior by extracting input features of a user power consumption curve.
5. The deep convolutional neural network-based power stealing monitoring system as claimed in claim 3, wherein the convolutional layer is composed of convolutional kernels with learning parameter function, which effectively extracts the input features, and during the forward propagation, the convolutional kernels use the input quantity to perform convolution calculation, and the convolutional kernels move on the result of data preprocessing on the historical monitoring data according to a predefined step length, namely the distance of each movement of the convolutional kernels, so as to retain the local features of the historical monitoring data and extract more abstract features.
6. The deep convolutional neural network-based power theft monitoring system of claim 3, wherein the pooling layer reduces the dimensionality of the features while preserving important information.
7. The deep convolutional neural network-based power stealing monitoring system of claim 3, wherein the fully-connected layer is composed of rearranged neurons at the end of the power stealing deep convolutional neural network model architecture.
8. The deep convolutional neural network-based power stealing monitoring system of claim 1, wherein the power stealing deep convolutional neural network model monitoring effect assessment module evaluates power stealing monitoring by adopting accuracy, recall rate, specificity, F1 score and accuracy index.
9. A power stealing monitoring method based on a deep convolutional neural network comprises the following steps:
s1, acquiring monitoring data;
s2, preprocessing the acquired data;
s3, inputting the preprocessed data into a power stealing detection deep convolution neural network model;
and S4, outputting a judgment result, and if the judgment result is consistent with the electricity stealing model data, giving an alarm to the data.
10. The method for monitoring power theft based on deep convolutional neural network of claim 9, wherein the inputting of the preprocessed data into the model of deep convolutional neural network for detecting power theft comprises the following steps:
s3-1, acquiring historical monitoring data;
s3-2, carrying out data preprocessing on the historical monitoring data;
s3-3, training a power stealing monitoring deep convolution neural network model until convergence;
and S3-4, evaluating the electricity stealing monitoring deep convolutional neural network model according to the confusion matrix evaluation index.
CN202010259498.9A 2020-04-03 2020-04-03 Electricity stealing monitoring system and method based on deep convolutional neural network Pending CN111461923A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010259498.9A CN111461923A (en) 2020-04-03 2020-04-03 Electricity stealing monitoring system and method based on deep convolutional neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010259498.9A CN111461923A (en) 2020-04-03 2020-04-03 Electricity stealing monitoring system and method based on deep convolutional neural network

Publications (1)

Publication Number Publication Date
CN111461923A true CN111461923A (en) 2020-07-28

Family

ID=71685905

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010259498.9A Pending CN111461923A (en) 2020-04-03 2020-04-03 Electricity stealing monitoring system and method based on deep convolutional neural network

Country Status (1)

Country Link
CN (1) CN111461923A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112098714A (en) * 2020-08-12 2020-12-18 国网江苏省电力有限公司南京供电分公司 ResNet-LSTM-based electricity stealing detection method and system
CN113408804A (en) * 2021-06-24 2021-09-17 广东电网有限责任公司 Electricity stealing behavior detection method, system, terminal equipment and storage medium
CN113469266A (en) * 2021-07-14 2021-10-01 广西电网有限责任公司 Electricity stealing behavior detection method based on improved deep convolutional neural network

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107966600A (en) * 2017-09-30 2018-04-27 国家电网公司 A kind of electricity anti-theft system and its electricity anti-theft method based on deep learning algorithm
CN109859428A (en) * 2019-03-12 2019-06-07 上海荷福人工智能科技（集团）有限公司 A kind of comprehensive theft management method and system based on artificial intelligence
CN110298574A (en) * 2019-06-21 2019-10-01 国网辽宁省电力有限公司鞍山供电公司 A kind of electricity consumption subscriber payment risk rating method based on convolutional neural networks
CN110705694A (en) * 2019-09-30 2020-01-17 国网山东省电力公司电力科学研究院 Electric larceny monitoring method facing edge data center based on feature extraction

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107966600A (en) * 2017-09-30 2018-04-27 国家电网公司 A kind of electricity anti-theft system and its electricity anti-theft method based on deep learning algorithm
CN109859428A (en) * 2019-03-12 2019-06-07 上海荷福人工智能科技（集团）有限公司 A kind of comprehensive theft management method and system based on artificial intelligence
CN110298574A (en) * 2019-06-21 2019-10-01 国网辽宁省电力有限公司鞍山供电公司 A kind of electricity consumption subscriber payment risk rating method based on convolutional neural networks
CN110705694A (en) * 2019-09-30 2020-01-17 国网山东省电力公司电力科学研究院 Electric larceny monitoring method facing edge data center based on feature extraction

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112098714A (en) * 2020-08-12 2020-12-18 国网江苏省电力有限公司南京供电分公司 ResNet-LSTM-based electricity stealing detection method and system
CN112098714B (en) * 2020-08-12 2023-04-18 国网江苏省电力有限公司南京供电分公司 Electricity stealing detection method and system based on ResNet-LSTM
CN113408804A (en) * 2021-06-24 2021-09-17 广东电网有限责任公司 Electricity stealing behavior detection method, system, terminal equipment and storage medium
CN113469266A (en) * 2021-07-14 2021-10-01 广西电网有限责任公司 Electricity stealing behavior detection method based on improved deep convolutional neural network
CN113469266B (en) * 2021-07-14 2022-08-23 广西电网有限责任公司 Electricity stealing behavior detection method based on improved deep convolutional neural network

Similar Documents

Publication Publication Date Title
CN112784881B (en) Network abnormal flow detection method, model and system
CN109302410B (en) Method and system for detecting abnormal behavior of internal user and computer storage medium
CN111461923A (en) Electricity stealing monitoring system and method based on deep convolutional neural network
CN111562108A (en) Rolling bearing intelligent fault diagnosis method based on CNN and FCMC
Liang et al. Multi-scale dynamic adaptive residual network for fault diagnosis
CN108399434B (en) Analysis and prediction method of high-dimensional time series data based on feature extraction
CN112308124B (en) Intelligent electricity larceny prevention method for electricity consumption information acquisition system
CN111340086A (en) Method, system, medium and terminal for processing label-free data
CN111343147A (en) Network attack detection device and method based on deep learning
CN114169374A (en) Cable-stayed bridge stay cable damage identification method and electronic equipment
Gad et al. Performance evaluation of predictive models for missing data imputation in weather data
CN113447759A (en) Multi-classification RVM power grid fault discrimination method and system
CN114863170A (en) Deep learning-based new energy vehicle battery spontaneous combustion early warning method and device
CN114118174A (en) Rolling bearing fault diagnosis method and system
CN113987910A (en) Method and device for identifying load of residents by coupling neural network and dynamic time planning
CN113822336A (en) Cloud hard disk fault prediction method, device and system and readable storage medium
CN115343676B (en) Feature optimization method for positioning technology of redundant substances in sealed electronic equipment
Irfan et al. Energy theft identification using AdaBoost Ensembler in the Smart Grids
Pozdnoukhov et al. Applying machine learning methods to avalanche forecasting
CN116561569A (en) Industrial power load identification method based on EO feature selection and AdaBoost algorithm
Nawaiseh et al. Financial Statement Audit using Support Vector Machines, Artificial Neural Networks and K-Nearest Neighbor: An Empirical Study of UK and Ireland
CN113935413A (en) Distribution network wave recording file waveform identification method based on convolutional neural network
CN113762151A (en) Fault data processing method and system and fault prediction method
CN114492830A (en) Deep learning model depolarization method and device based on individual discrimination example pair generation
Liu et al. Structure simplification of neural network for smile classification.

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination