Electric energy meter current overload prediction method and system

Abstract

The invention provides an electric energy meter current overload prediction method. The method comprises the following steps: acquiring current overload related power utilization data characteristics and user power utilization sample data; preprocessing the data to obtain power consumption data with time sequence characteristics; the method comprises the following steps: classifying power consumption data with time sequence characteristics according to different time granularities to obtain multi-level power consumption time sequence data, extracting key characteristics in a power consumption data sequence according to a set rule, and obtaining candidate Shapelet sequences of different time windows; based on the Shaplet sequence features and the power consumption features, constructing and training an obtained electric energy meter current overload prediction model by adopting a gating neural network, and performing electric energy meter current overload judgment; based on the current overload judgment result of the electric energy meter, integrating the current overload distribution condition of each time level, forming a formatted early warning message, sending the formatted early warning message, carrying out field check, and updating data; according to the invention, normal monitoring and efficient management of the current overload problem can be realized.

Description

technical field [0001] The invention belongs to the technical field of intelligent electricity consumption, and in particular relates to a current overload prediction method and system of an electric energy meter. Background technique [0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art. [0003] Current overload refers to the situation that the electric energy meter exceeds the rated current. There are many reasons for the current overload, such as the increase of the voltage at the power supply terminal, the increase in the number or power of the electrical equipment in the circuit, and the increase in the resistance caused by the aging of the wires. In recent years, with the development of the power industry, especially the continuous innovation of user electricity information collection technology, the early manual collection has gradually developed into the current int...

AI Technical Summary

Problems solved by technology

[0004] In the big data analysis module of the electricity consumption information collection system, for the prediction model of electric energy meter current overload, commonly used methods include logistic regression, random forest, BP neural network (Back Propagation Neural Network, BPNN), radial basis function (Radial Basis Function, RBF) neural network and other classification models, the traditional machine learning model requires feature engineering before training the classifier, the labor cost is high, and the electricity consumption data has certain periodicity and non-periodicity, it is difficult to manually filter features Unearth the hidden information in it

Deep learning models such as BPNN and RBF can mine the potential correlation information in the electricity consumption data to a certain extent, and obtain better classification results, but such models are difficult to mine the mutation characteristics and long-term dependence in the electricity consumption data of users. relation

Images