Detection Measurement Equipment Fault in Power distribution Using Long Short-Term Memory on Automatic Meter Reading

Abstract

This paper describes the methods for detecting meter measuring equipment faults, which fall under the category of non-Technical losses (NTL). It has happened in power distribution network. The history data record of the voltage and current from the Automatic Meter Reading (AMR) database is manipulated in this study. The obtained data is performed in a pattern CSV file and feature extracted by an electrical technician. Then, the dataset is fed into the Long Short-Term Memory (LSTM) model to distinguish the event type including normal, voltage fault, current fault, and communication. This model provided accuracy detection to achieve 99%. The model can find out the problem quickly and accurately and the electricity company can solve the problem suddenly and help reduce NTL in power distribution.