Internal Leakage Diagnosis of a Hydraulic Cylinder Based on Optimization DBN Using the CEEMDAN Technique

Table 3

Optimization parameters of the DBN.

Parameters	Value
Number of nodes in the input layer	2000
Number of nodes in the output layer	4
Number of RBMs	3
Number of iterations for each RBM	200
Number of nodes in the first hidden layer (obtained using the PSO-SA)	860
Number of nodes in the second hidden layer (obtained using the PSO-SA)	284
Number of nodes in the third hidden layer (obtained using the PSO-SA)	78
Learning rate of the deep belief network (obtained using the PSO-SA)	0.48
Momentum of the deep belief network (obtained using the PSO-SA)	0.23