CN-121981381-A - Metering point abnormity monitoring and analyzing method based on electricity consumption

CN121981381ACN 121981381 ACN121981381 ACN 121981381ACN-121981381-A

Abstract

The metering point anomaly monitoring analysis method based on the electricity consumption information comprises the steps of adaptively learning legal electricity consumption time sequence data and illegal electricity consumption time sequence data based on a domain adaptive neural network to generate mine adaptive anomaly feature vectors, processing the mine adaptive anomaly feature vectors and the legal electricity consumption time sequence data by adopting a generation countermeasure network to obtain generation samples, forming an anomaly sample set by the illegal electricity consumption time sequence data, constructing an anomaly monitoring analysis model, training the anomaly monitoring analysis model by taking the anomaly sample set and the legal electricity consumption time sequence data as training samples, collecting real-time electricity consumption time sequence data of metering points, inputting the real-time electricity consumption time sequence data into the anomaly monitoring analysis model to obtain illegal mining risk scores, supplementing the sparse illegal electricity consumption data by means of field transition and generation countermeasure, increasing the training samples of the anomaly monitoring analysis model, improving the training effect, and facilitating accurate monitoring of the illegal electricity consumption behavior of the mine.

Inventors

YANG JUN
YAO XUESONG
ZHANG YONG
LIU YALI
SHI BINBIN
LIANG BO
YANG PEI
WANG JIANHUI
Qiao Kehua
REN XIANG

Assignees

临汾汾能电力科技试验有限公司
国网山西省电力有限公司临汾供电分公司

Dates

Publication Date: 20260505
Application Date: 20251231

Claims (10)

1. The metering point abnormity monitoring and analyzing method based on the electricity consumption information is characterized by comprising the following steps of: Step S1, legal electricity utilization time sequence data, illegal electricity utilization time sequence data and illegal electricity utilization time sequence data of illegal scenes in the similar field are obtained; step S2, performing self-adaptive learning on legal electricity utilization time sequence data and illegal electricity utilization time sequence data based on a field self-adaptive neural network, and generating mine adaptation abnormal feature vectors; s3, processing the mine adaptation abnormal feature vector and the legal electricity utilization time sequence data by adopting a generation countermeasure network to obtain a generation sample, and forming an abnormal sample set by the generation sample and the illegal electricity utilization time sequence data; S4, constructing an anomaly monitoring analysis model, and training the anomaly monitoring analysis model by taking an anomaly sample set and legal electricity utilization time sequence data as training samples; And S5, collecting real-time electricity utilization time sequence data of the metering point, and inputting the real-time electricity utilization time sequence data into an anomaly monitoring analysis model to obtain illegal mining risk scores.
2. The electricity consumption information-based metering point anomaly monitoring and analyzing method according to claim 1, wherein the specific steps of the step S1 include: S11, legal electricity utilization time sequence data of different seasons, different equipment working conditions and different policy stages of the mine are extracted from an enterprise production record database of the mine; step S12, acquiring illegal power utilization time sequence data of marked illegal mining of the mine from a published archive database; step S13, screening illegal electricity utilization time sequence data of illegal scenes in similar fields including high-energy consumption small workshops, illegal sand yards and illegal smelting plants from the power company client files; And step S14, preprocessing the illegal power utilization time sequence data, the illegal power utilization time sequence data and the illegal power utilization time sequence data.
3. The electricity consumption information based metering point anomaly monitoring and analyzing method according to claim 2, wherein the preprocessing of step S14 includes data cleansing, unified time stamping, unified electricity consumption units and standardized processing.
4. The electricity consumption information-based metering point anomaly monitoring and analyzing method according to claim 1, wherein the specific steps of the step S2 include: step S21, respectively extracting features of legal power utilization time sequence data and illegal power utilization time sequence data to obtain legal features and illegal features; S22, inputting legal features and illegal features into a domain self-adaptive neural network, performing cross-domain migration by the domain self-adaptive neural network, and generating mine adaptation abnormal feature vectors of the attached mine; and S23, carrying out confidence degree scoring on the generated mine adaptation abnormal feature vector, and eliminating the mine adaptation abnormal feature vector with the confidence degree score smaller than a preset threshold value.
5. The electricity consumption information-based metering point anomaly monitoring and analyzing method according to claim 4, wherein the specific steps of step S22 include: Step S221, stripping the domain-specific features of legal features and illegal features through countermeasure training; step S222, carrying out maximum mean value difference minimization on the stripped legal features and the stripped illegal features; And S223, expressing abnormal electricity characteristics irrelevant to the learning field, and generating mine adaptation abnormal characteristic vectors attached to the mine.
6. The electricity consumption information-based metering point anomaly monitoring and analyzing method according to claim 1, wherein the specific steps of the step S3 include: Step S31, carrying out feature stitching on the mine adaptation abnormal feature vector and legal electricity utilization time sequence data to obtain a stitched feature vector; step S32, inputting the spliced characteristic vector into a generator for generating an countermeasure network, and generating simulated electricity time sequence data of illegal mining of the mine by the generator based on the physical constraint of electricity consumption of the mine; Step S33, judging the simulated electricity time sequence data by a discriminator for generating an countermeasure network, and screening to obtain a generated sample; step S34, combining the simulated power consumption time series data and the illegal power consumption time series data in the generated samples into an abnormal sample set.
7. The electricity consumption information-based metering point anomaly monitoring analysis method according to claim 6, wherein the discriminator comprises an authenticity discriminating branch for discriminating real mine data from a generated sample and an anomaly discriminating branch for discriminating normal electricity consumption from abnormal electricity consumption, and the generated sample is determined to be abnormal by the screening discriminator and has a confidence level larger than a preset threshold.
8. The electricity consumption information-based metering point anomaly monitoring and analyzing method according to claim 1, wherein the specific steps of the step S4 include: S41, constructing an LSTM-Attention lightweight anomaly monitoring analysis model, wherein the anomaly monitoring analysis model comprises an input layer, an LSTM hidden layer, an Attention layer and an output layer; s42, dividing the abnormal sample set and legal electricity time sequence data into a training set, a verification set and a test set, and inputting the training set into an abnormal monitoring analysis model; s43, training an anomaly monitoring analysis model, and reinforcing the weight of key anomaly characteristics in a non-production period through an attribute layer; And S44, adjusting the super parameters of the abnormal monitoring analysis model by using the verification set, verifying the performance of the abnormal monitoring analysis model by using the test set, and completing training when the accuracy reaches a preset threshold.
9. The electricity consumption information-based metering point anomaly monitoring and analyzing method according to claim 8, wherein the specific steps of step S5 include: step S51, collecting real-time power utilization time sequence data from mine metering points, and performing data cleaning, abnormal value elimination and missing data complementation on the real-time power utilization time sequence data; S52, extracting time sequence characteristics of the real-time power utilization time sequence data, forming characteristic vectors, and inputting the characteristic vectors into an anomaly monitoring analysis model; and S53, processing the feature vector by the anomaly monitoring analysis model, and outputting an illegal mining risk score of 0-1 score.
10. The electricity consumption information based metering point anomaly monitoring analysis method of claim 9, further comprising: Acquiring real-time electricity utilization time sequence data which is recently output by an anomaly monitoring analysis model and has an illegal mining risk of 0 score as recent electricity utilization time sequence data; respectively extracting characteristic space distribution of the recent power utilization time sequence data and legal power utilization time sequence data and calculating cosine similarity; When the cosine similarity is smaller than a preset similarity threshold, judging that the model drifts, freezing an input layer and an LSTM hidden layer of the anomaly monitoring analysis model, and adjusting weights of an attribute layer and an output layer.

Description

Metering point abnormity monitoring and analyzing method based on electricity consumption Technical Field The invention relates to the technical field of electricity consumption monitoring, in particular to a metering point abnormity monitoring and analyzing method based on electricity consumption. Background The illegal mining of the mine refers to the mining behaviors of violating the laws and regulations of mineral resources, such as unauthorized mining without taking mining licenses, mining beyond the specified range of mining licenses, and the like, the illegal mining behaviors are often accompanied by unordered mining, excessive mining and illegal occupation and abuse of electric power resources, which not only cause interference to the normal operation of a normal mineral enterprise, but also threaten the safety and the stability of electric power supply, in addition, the illegal mining activities can possibly cause secondary disasters such as geological disasters, environmental pollution and the like, and seriously affect the life and property safety and social stability of people, and along with the continuous concealment and the technological miniaturization of the illegal mining methods, the traditional manpower inspection supervision means has hardly met the requirements of the current situation, so that the digital technology is urgently needed to take intelligent electricity consumption monitoring means for the mine enterprise to strengthen the supervision and fight the illegal mining behaviors, ensure the safety and stability of the development order of the mineral resources, but the abnormal monitoring of the metering points based on the electric information becomes an effective means of illegal mining, but the nature of the illegal mining behaviors causes serious influence on the life and property safety and social stability of the people, and extremely difficult to support the training model of the illegal electricity consumption time series supervision. Disclosure of Invention In view of the above, the invention provides a metering point anomaly monitoring analysis method based on electricity consumption information, which can be used for training a model by expanding the anomaly electricity consumption information in the similar field, so that the method can be accurately applied to anomaly monitoring analysis of mine metering points. The technical scheme of the invention is realized as follows: The metering point abnormity monitoring and analyzing method based on the electricity consumption information comprises the following steps: Step S1, legal electricity utilization time sequence data, illegal electricity utilization time sequence data and illegal electricity utilization time sequence data of illegal scenes in the similar field are obtained; step S2, performing self-adaptive learning on legal electricity utilization time sequence data and illegal electricity utilization time sequence data based on a field self-adaptive neural network, and generating mine adaptation abnormal feature vectors; s3, processing the mine adaptation abnormal feature vector and the legal electricity utilization time sequence data by adopting a generation countermeasure network to obtain a generation sample, and forming an abnormal sample set by the generation sample and the illegal electricity utilization time sequence data; S4, constructing an anomaly monitoring analysis model, and training the anomaly monitoring analysis model by taking an anomaly sample set and legal electricity utilization time sequence data as training samples; And S5, collecting real-time electricity utilization time sequence data of the metering point, and inputting the real-time electricity utilization time sequence data into an anomaly monitoring analysis model to obtain illegal mining risk scores. Preferably, the specific steps of the step S1 include: S11, legal electricity utilization time sequence data of different seasons, different equipment working conditions and different policy stages of the mine are extracted from an enterprise production record database of the mine; step S12, acquiring illegal power utilization time sequence data of marked illegal mining of the mine from a published archive database; step S13, screening illegal electricity utilization time sequence data of illegal scenes in similar fields including high-energy consumption small workshops, illegal sand yards and illegal smelting plants from the power company client files; And step S14, preprocessing the illegal power utilization time sequence data, the illegal power utilization time sequence data and the illegal power utilization time sequence data. Preferably, the preprocessing in step S14 includes data cleansing, uniform time stamping, uniform electricity usage units, and standardized processing. Preferably, the specific steps of the step S2 include: step S21, respectively extracting features of legal power utilization time sequence data and illegal powe