User power consumption data anomaly detection method and device and computer equipment

Abstract

The invention relates to a user power consumption data anomaly detection method and device, computer equipment and a storage medium. The method comprises the following steps: acquiring historical power consumption data of a user and real power consumption data at the current moment; inputting the historical power consumption data into a trained anomaly detection model, determining a first Boltzmann distribution feature of the historical user data, and outputting predicted power consumption data at the current moment according to the first Boltzmann distribution feature; calculating according to the predicted power consumption data and the real power consumption data to obtain the distance between the predicted power consumption data and the real power consumption data; and when the distance is greater than a preset threshold value, marking the real power consumption data as abnormal power consumption data. By adopting the method, the accuracy of abnormal power consumption data identification can be improved.

Description

technical field [0001] The present application relates to the technical field of user electricity data processing, and in particular to a method, device, computer equipment and storage medium for abnormal detection of user electricity data. Background technique [0002] With the development of user electricity data processing technology, due to various reasons such as electricity theft and meter failure, user electricity consumption data will be abnormal. Analyzing and judging, there is an abnormal detection technology for user electricity data. [0003] In the traditional technology, from the perspective of the timing of electricity consumption data, the proposed methods can be roughly divided into two categories: unsupervised learning methods and supervised learning methods. The methods based on unsupervised learning mainly include cluster analysis, time series clustering, etc. The key to this type of method is to define a distance metric to model the dissimilarity betwe...

AI Technical Summary

Problems solved by technology

[0004] However, in the traditional method, the timing characteristics of power consumption in different industries in unsupervised learning are quite different. Without reasonable quantification of the characteristics of power consumption of each user, it is difficult to determine whether the large difference is due to abnormal power consumption of users or due to industrial consumption. It is caused by the characteristics of electricity, so it is easy to cause wrong identification of abnormal electricity consumption data

Images