A Smart Grid Intrusion Detection System Based on Optimization

Abstract

Smart grid significantly improves the functions of conventional power networks, but it also makes them more susceptible to attacks of different types. Using smart grid’s vulnerabilities, attackers may compromise the integrity and confidentiality of networks, and obtain access to them as a result. Intrusion Detection System (IDS) constitutes an important means to provide safe and reliable services in a smart grid environment. This paper proposes an intrusion detection model under smart grid environment that is widely distributed in the three-tier architecture of power grid system. The model introduces Random Forest (RF) into machine learning and Particle Swarm Optimization (PSO) into evolutionary computation. To improve the accuracy of the model, this paper taps into the adaptive strategy for continued adjustment concerning the flight attitude of particles, and uses multiple mutation methods to prevent the occurrence of local optimum. The AMPSO algorithm proposed in this paper performs well in benchmarking function tests, thereby providing effective guarantee for the improvement of RF classifier. According to simulation experiments based on test sets of NPL-KDD, the intrusion detection model proposed in this paper is highly effective and practical.