TEA-EKHO-IDS: An intrusion detection system for industrial CPS with trustworthy explainable AI and enhanced krill herd optimization

Industrial cyber-physical systems (CPS) are vulnerable to cyberattacks that can compromise their operations, safety, and security. Traditional intrusion detection systems are ineffective in detecting and preventing cyberattacks in industrial CPS due to their dynamic and complex nature. In addition, there is a lack of interpretability and transparency in the decision-making process of IDS, which makes it challenging for system administrators to identify risks. Furthermore, deep learning approaches can be computationally intensive and challenging to deploy and scale in industrial CPS. This study proposes a novel intrusion detection system called TEA-EKHO-IDS that utilises trustworthy explainable artificial intelligence (XAI) and enhanced krill herd optimisation (EKHO) to detect breaches in the CPS. The proposed method uses XAI-EKHO for feature selection, which gives more robust global searching capabilities and faster convergence time by estimating the decision weighting factor. The intrusion detection performance is optimised through the integration of explainable AI, bi-directional LSTM, and Bayesian optimisation (BO-Bi-LSTM) for efficient detection and classification. The proposed approach has been shown to effectively and accurately identify and classify intrusions with a success rate of 98.96%, as demonstrated through experiments and analysis results. The TEA-EKHO-IDS system provides a promising solution for detecting cyber-attacks in industrial CPS, making it a valuable addition to existing industrial security systems. Graphical abstract