Equipment, device and method for predicting electricity stealing probability of users on basis of trust region algorithm

Abstract

An embodiment of the invention discloses equipment, a device and a method for predicting an electricity stealing probability of users on the basis of trust region algorithm. The method includes: acquiring characteristic factors and electricity stealing information of historical users; constructing an electricity stealing probability prediction function according to a logistic regression model and the characteristic factors of the historical users, wherein the electricity stealing probability prediction function includes a to-be-determined coefficient vector; transforming the prediction fitting function to obtain a cost function according to the characteristic factors and the electricity stealing information of the historical users; solving the cost function on the basis of the trust region algorithm to obtain a value of the coefficient factor; acquiring the electricity stealing probability prediction function according to the value of the coefficient factor; determining the electricity stealing probability of current users according to characteristic factors of the current users and the electricity stealing probability prediction function. By adoption of the technical scheme, efficiency and accuracy in prediction of the electricity stealing probability of the users are improved.

Description

technical field [0001] The embodiments of the present invention relate to a probability prediction method, in particular to a method, device and equipment for predicting the probability of stealing electricity by a user based on a trust region algorithm. Background technique [0002] Abnormal electricity consumption by users is a prominent problem that plagues the normal supply of electricity and the healthy development of power supply companies. Many external behaviors that cause abnormal electricity consumption, especially electricity theft, have brought a great negative impact on national property and corporate income, disrupting the normal Power order. Taking Guangdong as an example, in 2014, a total of 3,964 electricity theft cases were investigated and dealt with in Guangdong Province, with a total loss of 52.2312 million yuan. [0003] Due to the objective existence of various factors such as frequent load adjustment of power supply lines, users' handling of power co...

AI Technical Summary

Problems solved by technology

However, user electricity consumption characteristic data involves more data, more data types, large volume, and fast increment. In the process of data mining, there are high requirements for the efficiency of algorithms.

Images