With the constant development of drilling industry,the design of the top drive device has tended to be mature.However,there is still a lack of research on condi-tion monitoring and intelligent diagnosis for the top drive rotating system.Generally a pas-sive policy of breakdown/periodic maintenance is adopted which greatly increases the cost of maintenance and reduces the benefit of the enterprise.This paper,aiming at this phe-nomenon,proposes the deep Boltzmann machine model (DBM) preprocessed by Gauss fil-ter.Then optimize the G-DBM model with non-sampling lifting wavelet packet and Particle Swarm Optimization algorithm.Through the practical case analysis,the fault conditions of the determined positions in the system were diagnosed and the correct rate reached 97.97%,which verified the effectiveness of the method proposed in this paper,providing powerful technical support for intelligent diagnosis and fault identification,further more a basis for making intelligent diagnostic maintenance strategies.