UTILIZING A UNIQUE DEEP LEARNING TECHNIQUE FOR DETECTING ANOMALIES IN INDUSTRIAL AUTOMATION SYSTEMS

Abstract

Industrial automation systems (IASs) are utilized in vital facilities to sustain society's fundamental services. As a consequence, protecting them against terrorist operations, natural catastrophes and cyber-threats is essential. The research on techniques for identifying cyber-attacks in IAS environments is lacking. The study proposed the Stochastic Turbulent water flow optimization based restricted Boltzmann machine (STWFO-RBM) to overcome the challenges. The proposed STWFO-RBM integrates anomaly detection into the fabric of industrial automation, enhancing system resilience and responsiveness. We collected datasets from the water industry and preprocessed them through min-max normalization, and then principal component analysis was used for feature extraction. The results show that the suggested technique applies to a real-world IAS situation, with state-of-the-art accuracy of 97%, F1 score of 96%, precision of 98%, recall of 95% and 6.1s of computational time. Our proposed method is better than the average of earlier endeavors.

Keywords

• industry
• stochastic turbulent water flow optimization based restricted boltzmann machine (stwfo-rbm)
• anomalies
• real-time monitoring.