Objective Machine learning algorithms, with their powerful capabilities in data processing, nonlinear fitting, and adaptive optimization, provide crucial support for the efficient operation and intelligent evolution of compressed air energy storage (CAES).

Method This paper reviews the advances of machine learning in the field of CAES, systematically summarizing the technical approaches and practical achievements across various application domains. It begins by briefly outlining the system architecture and core equipment of CAES plants, as well as the typical workflow of machine learning algorithms. The review then delves into specific applications of machine learning, including system modeling and performance prediction, condition monitoring and fault diagnosis, intelligent control and operational optimization, and its use in underground cavern management.

Result The integration of machine learning with CAES has successfully enabled performance prediction for key equipment such as compressors, expanders, and heat exchangers. Significant breakthroughs have also been achieved in equipment condition diagnosis and system parameter optimization, markedly enhancing the operational intelligence of CAES systems.

Conclusion In the future, with the growing number of commercial CAES plants and the continuous iteration of machine learning algorithms, this synergy will further drive the evolution of CAES facilities into intelligent power plants featuring autonomous perception, adaptive regulation, and smart decision-making. This advancement will provide core support for the flexible operation of future power systems.