Aimed at the problem that the decrease in population diversity of the Sparrow Search Algorithm (SSA) in the later iterations is easy to lead the local extreme value solution of the algorithm,the t distribution mutation strategy was introduced to improve the SSA,and the improved SSA was used to optimize the parameters of Variational Mode Decomposition(VMD),so as to obtain the optimal parameter combination of the VMD algorithm.Combining the improved SSA-VMD algorithm with refined composite multi-scale fuzzy entropy (RCMFE),the feature extraction of reciprocating compressor bearing clearance faults was carried out.The experimental results show that compared with the original SSA-VMD algorithm,the improved SSA-VMD algorithm has better separability in RCMFE feature analysis and higher fault recognition accuracy.