Prognostics for the Sustainability of Industrial Cyber-Physical Systems: From an Artificial Intelligence Perspective

As industrial cyber-physical systems (ICPS) play an increasingly pivotal role in the new industrial paradigm, their sustainability has become the current research focus. Remaining useful life (RUL) prediction, also known as prognostics, is critically significant for the sustainability of ICPS. The prognostics involve utilizing process monitoring devices within ICPS to acquire real-time operational data. Based on the trends observed in the monitored data, the analysis predicts when potential system failures may occur. Accurate prognostics allow real-time monitoring of the system's health status, which enables early warning of possible faults and system reliability. The primary objective of this paper is to provide readers with a timely survey and review that reveals the current research status, development trends, and common challenges in the prognostics domain within ICPS. From the perspective of artificial intelligence (AI), the paper comprehensively reviews predictive approaches based on stochastic process, machine learning, and their hybrid applications. Through a comprehensive comparison of existing approaches, the paper delves into the strengths and weaknesses of these approaches. Furthermore, facing some cutting-edge issues in existing RUL prediction approaches for ICPS, this paper analyzes some pioneering investigations that have achieved great results. Finally, the paper explores the opportunities and challenges of prognostics from the perspective of artificial intelligence, which aims to drive the sustainability of ICPS.