Machine learning applications in Cyber-Physical Production Systems: a survey

Abstract

Cyber-Physical Production Systems (CPPS) play a vital role in realizing the vision of Industry 4.0. In the last decade, various machine learning methods have been implemented in manufacturing systems to improve their intelligence. However, few review papers on machine learning applications in CPPS have been published. In this context, this paper presents a survey of machine learning applications in Cyber-Physical Production Systems. Both bibliometric analysis and qualitative analysis have been conducted based on the related literatures published in the last decade. We identified the major research issues with respect to machine learning applications in CPPS, i.e. anomaly detection, predictive maintenance, fault management, efficiency, quality assurance, and scheduling. The review results show that although machine learning has been extensively applied in manufacturing, its applications in CPPS have not been widely studied. Based on the detailed discussions of the research issues and challenges, this paper indicates the current limitations of CPPS and demonstrates the great advantages and potential for applying machine learning in CPPS in future research.