Bearing Fault Diagnosis Based on Deep Belief Network and Multisensor Information Fusion

Table 1

Procedure for bearing fault diagnosis using multi-information with DBN.

Step	Description
Step	Gather multichannel vibration signals
Step	Extract features of each channel sample
Step	Input all features of training samples and initial parameters of the DBN
Step	Optimize the DBN structure using reconstruction error of multi-information fusion
	() Each layer of the DBN is trained using RBM learning rule
	() Fine-tune the DBN using backpropagation learning
	() Calculate the reconstruction error using model outputs and the target label data
	() If the reconstruction error is smaller than ε output the DBN structure; otherwise, and return to step ()
	() If the reconstruction error is not smaller than and , and return to step () until is more than
Step	Develop the DBN using the structure with the best reconstruction error
Step	Perform diagnosis using the training DBN classifier model