A Review on Detection of Cyberattacks in Industrial Automation Systems and Its Advancement Through MPC-Based AI

Abstract

The burgeoning digitalization of industrial automation systems (IASs) has amplified their vulnerability to sophisticated cyberattacks, necessitating robust and adaptable detection mechanisms. This review provides insights into the current landscape of cyberattack detection in IASs and underscores the potential of model predictive control (MPC)-based AI techniques to bolster security and resilience in critical infrastructure settings. By harnessing the combined strengths of physically grounded MPC models and data-driven AI algorithms, this framework offers significant advantages over traditional methods. The review navigates through existing literature, scrutinizing diverse approaches for cyberthreat detection. An emphasis is placed on the proactive nature of MPC, which enables the modeling and optimization of complex system dynamics, coupled with the adaptability and learning capabilities of AI algorithms. These features collectively empower the system to identify anomalies indicative of cyberattacks in real-time, thus fortifying IAS against potential disruptions. Key findings from reviewed studies demonstrate the previous technologies in detecting and mitigating cyberthreats while maintaining the stability and functionality of industrial processes. The review further highlights a problem formulation based on MPC method to detect cyberattacks, including computational efficiency and real-time responsiveness. This review concludes by affirming the immense potential of MPC-based AI to revolutionize cyberattack detection in IAS. Its robust and adaptable nature offers a compelling alternative to existing methods, paving the way for securing critical infrastructure in the face of ever-evolving cyberthreats.