Abstract: Composite materials have wide applications in many fields due to their excellent comprehensive properties. However, as the complexity of the components, structures, and performance requirements of composite materials has increased, the traditional research methods in the composite materials field, such as experimental studies and computational simulations, face problems such as high cost, long research cycle, high data requirements, complex modeling and poor interpretability. Machine learning (Machine Learning, ML) methods, as an artificial intelligence technology, have the ability of automatic learning, high-dimensional data processing, analysis and prediction, and classification and decision-making, which can effectively solve the problems existing in traditional composite materials research methods and are considered a new technology in composite materials structure design, analysis, and prediction. It has become a trend in the field of composite materials research. This paper reviews and evaluates the latest research achievements of ML methods applied to the composite materials field, focusing on the research progress in composite material mechanical property prediction, structure optimization design, and damage detection, and discusses and prospects its future development direction.