A Lightweight Intrusion Detection and Electricity Theft Detection System for Smart Grid

Smart grid systems have improved networking for power systems and many other industrial systems, but they still have many vulnerabilities, making them an easy target for cyber attacks. Recently, the number of attacks has also increased. The present work investigates the reliability and security of Smart Grid (SG). The reliability and security are investigated in two aspects that are electricity fraud detection followed by the intrusion detection system. This work presents the lightweight Intrusion detection system for SCADA and Modbus-based control systems that can detect intrusion with very high accuracy. The IDS developed is based on the ICS (industrial control system) dataset, which has 20 features (column) and 2,74,628 rows. The IDS dataset contains the Modbus packet’s attributes and network and physical infrastructure attributes. The IDS work is followed by detecting electricity theft on a realistic electricity consumption dataset released by the State Grid Corporation of China. A total of 42,372 users’ power usage data from 1,035 days is included in the data collection (from 1 January 2014 to 31 October 2016). Eight classifiers, as well as two basic neural networks (1DCNN and ANN), have been investigated on this dataset.