A Safe and Reliable Bayesian Spiking Neural Network in Industrial Control Systems
Tala Talaei Khoei, Naima Kaabouch · 2024
Industrial Control Systems refer to a combination of hardware, software, and network systems used to control and monitor industrial processes in sectors such as manufacturing, energy, and transportation. Ensuring the safety, effectiveness, and reliability of these systems is essential to prevent accidents, maintain operational efficiency, and protect critical infrastructure. To avoid these incidents, several methods based on Artificial Intelligence have been proposed in literature; however, most of these techniques deal with low robustness, limited safety and reliability, and low adaptability. In this paper, we propose Bayesian Spiking Neural Network to detect and classify abnormal behavior in Industrial Control Systems using the CASPER dataset. The evaluation was performed using several metrics. These metrics are the accuracy, probability of detection, probability of misdetection, probability of false alarm, aleatoric uncertainty, epistemic uncertainty, safety score, processing time, training time per sample, and memory usage. The results prove the high efficiency, safety, and reliability of the proposed technique.