A comprehensive knowledge map for AI improving security management of cyber-physical system enabled smart manufacturing

The enhancement of security management for smart manufacturing has garnered considerable attention, prompting extensive research efforts that explore AI-based secure Cyber-Physical System (CPS) methods. While previous studies have made valuable contributions to this field, the current state of literature reviews still requires a comprehensive examination to construct the knowledge map for future research. This study employed scientometric analyses, including collaboration network analysis, co-occurrence network analysis, and cocitation network analysis, to examine 1133 previous articles conducted between 2015 and 2023 concerning the research topic of AI application in secure CPS-enabled smart manufacturing. The findings reveal that the academic communities in this field comprise influential authors affiliated with leading academic institutions in China, India, and the USA. The key research topics of AI application in secure CPS-enabled smart manufacturing show evolutionary trends during different research periods. Existing research on AI application in secure CPSenabled smart manufacturing is supported by classical references, which can be divided into four significant co-citation clusters, including the theory of security and privacy in CPS, DL-based IDS, DDoS attacks on CPS by botnet, and FL method for privacy-preserved data sharing. Finally, this study predicted four potential future research directions for the research field of AI application in secure CPS-enabled smart manufacturing: the wellstructured and representative dataset for model training, methods for resisting PUF attacks against ML techniques, DL-based methods for detecting and resisting botnets, and lightweight FL-based methods for privacypreserved data sharing. The findings of this study can guide researchers in this field for future collaboration and work.