HBF-PSO and HNA-NN Based Intrusion Detection System for SCADA Networks

Abstract

The increasing adoption of remote-controlled, self-contained production machines has led to the integration of Supervisory Control and Data Acquisition (SCADA) systems as a key component of industrial automation. While machine connectivity has improved productivity, the threat of cybersecurity attacks has introduced weaknesses into control systems. This article proposes the development of an intrusion detection system (IDS) that optimizes search efficiency and diversity in the search population by implementing the Hummingbird Flight-based Particle Swarm Optimization (HBF-PSO) algorithm combined with the Hierarchical Neuron Architecture Neural Network (HNA-NN). The HBF strategy models incremental, energy-efficient flight patterns to improve feature optimization, while the HNA-NN classifier categorizes attack attempts with high precision. Experiments conducted on actual SCADA system databases (MORD, MIRD, SORD, and SIRD) have confirmed the efficiency of the proposed system, with 98.12% detection accuracy and 100% precision in the SORD database. The false-positive rate of the proposed system was 0% in both the MORD and SIRD databases. In general, the hybrid model has shown improved detection accuracy and specificity compared to traditional systems.