A multi-objective locust optimization algorithm (MOGOA) optimized variational mode decomposition (VMD) and least squares support vector machine (LSSVM) fault diagnosis method is proposed to address the difficulty in accu-rately diagnosing bearing faults in coal based activated carbon production equipment. Firstly, the MOGOA algorithm is introduced to optimize the penalty factor and decomposition level of VMD by introducing a composite fitness function based on permutation entropy and kurtosis reciprocal normalization. Secondly, use optimized VMD de-composition to extract bearing vibration signals and select sensitive variational mode components (IMF) for recon-struction. Finally, by optimizing the LSSVM model through MOGOA, a MOGOA-LSSVM fault diagnosis model is formed. Compared with the GOA-LSSVM method, verify the advantages of the proposed method in fault diagnosis.