CNN-LSTM based Electricity Theft Detector in Advanced Metering Infrastructure
Rutuja Umesh Madhure, Radha Raman, Sandeep Kumar Singh
Abstract
In a smart grid, estimating the power requirements of various regions and detecting malicious practices is very crucial. Advanced Metering Infrastructure (AMI) is a key component of the smart grid system that uses smart meters. Due to the vulnerability of the smart meter against cyber attacks, a strong defense algorithm is needed. In this paper, a CNN-LSTM based deep learning methodology is proposed for power consumption forecasting and anomaly detection using the combination of Convolutional layers and Stacked Long Short Term Memory (LSTM) layers architecture. The performance of the proposed method is tested using a real smart meter dataset. Test results show that the proposed CNN-LSTM based method is able to detect cyber attacks with a high detection rate. The proposed method outperforms some of the previously existing methods in the literature on comparing the results.