Rolling bearing fault diagnosis based on intelligent optimized self-adaptive deep belief network

Abstract

Due to the structure of the rolling bearing itself and the complexity of its operating environment, the collected vibration signals tend to show strong non-stationary and time-varying characteristics. It has been a challenge to extract useful fault signature information from actual bearing vibration signals and identify bearing faults in the field of machinery in recent years. Therefore, this paper proposes a novel optimized self-adaptive deep belief network (DBN). The DBN is pre-trained by a minimum batch stochastic gradient descent, and then a back-propagation neural network and conjugate gradient descent are used to supervise and fine-tune the entire DBN model, which effectively improves the classification accuracy of the DBN. A salp swarm algorithm is used to optimize the DBN, and then the experience of the DBN structure is summarized. The vibration signal of the rolling bearing is analyzed by this method and it is confirmed that it has higher diagnostic accuracy and better convergence.