Abstract: This study presents an advanced Intrusion Detection System (IDS) optimized for securing Cyber-Physical Systems (CPS) through the application of machine learning and AI-based optimization techniques. The numerical case study conducted highlights the effectiveness of the IDS, achieving a high detection rate to accurately identify intrusions while maintaining a low false positive rate, ensuring minimal misclassification of normal activities. The system demonstrates resource efficiency by adhering to computational constraints, achieving a cost of C_c=1.2 GFLOPS, which is critical for CPS environments with limited computational resources. The use of Particle Swarm Optimization (PSO) effectively tunes the IDS parameters, enabling the system to balance multiple objectives, such as maximizing detection accuracy, minimizing false positives, and optimizing computational overhead. This approach not only ensures robust intrusion detection but also provides a scalable and adaptable framework for real-world CPS applications. By integrating machine learning and AI-driven optimization, the study offers a practical solution for enhancing the resilience of CPS against evolving cyber threats, paving the way for secure and efficient system operations in critical infrastructures.

Keywords: Optimization techniques, cyber-physical systems, intrusion detection systems, machine learning, AI-based optimization, Particle Swarm Optimization

| DOI: 10.17148/IARJSET.2025.12733