This comprehensive review explores recent advances in propulsion shaft fault diagnosis methods based on correlation entropy and GMSVM. We analyze various approaches to optimize fault detection and classification in response to mechanical disturbances. The paper discusses adaptive monitoring mechanisms, real-time analysis systems, and integration with predictive maintenance frameworks. Furthermore, we examine case studies from different applications and provide a comparative analysis of various implementation strategies. The findings suggest that properly configured diagnosis methods can significantly enhance system reliability while reducing operational costs.