Improved Intrusion Detection in Cyber-Physical Systems with Explainable AI and Hybrid Optimization

A network of physical and cyber components that exchange feedback with one another is known as a cyber-physical system (CPS). A CPS is necessary for day-to-day operations and authorizes vital infrastructure as it serves as the foundation for cutting-edge smart devices. Robust intrusion detection strategies for CPS settings have been developed in part because of recent developments in explainable artificial intelligence (XAI). The XAI-enabled intrusion detection method in secure cyber-physical systems (XAIID-SCPS) is developed in this work. Detecting and categorizing intrusions on a CPS platform is the primary focus of the suggested XAIID-SCPS approach. A Hybrid Enhanced Glowworm Swarm Optimization (HEGSO) algorithm is used to choose which features to use in the XAIID-SCPS method. With an Enhanced Fruitfly Optimization (EFFO) method for parameter standardization, an Improved Elman Neural Network (IENN) design was used to find intrusions. The XAIID-SCPS method also incorporates the XAI approach and Local interpretable model-agnostic explanation (LIME) to make the black-box method easier to understand and explain. This makes it possible to accurately define attacks. There is a 98.88% chance that the XAIID-SCPS technique will work better than other methods, as shown by the higher simulation numbers.