Research Article

Remaining Useful Life Prediction of Rolling Bearings Based on Multiscale Convolutional Neural Network with Integrated Dilated Convolution Blocks

Table 2

Parameters of MS-CNN.

LayerParameter

1st fused layerKS = 16 × 1; r = 1; n = 4; s = 1
KS = 16 × 1; r = 2; n = 4; s = 1
KS = 16 × 1; r = 3; n = 4; s = 1
2nd fused layerKS = 8 × 1; r = 1; n = 8; s = 1
KS = 8 × 1; r = 2; n = 8; s = 1
KS = 8 × 1; r = 3; n = 8; s = 1
3rd fused layerKS = 4 × 1; r = 1; n = 16; s = 1
KS = 4 × 1; r = 2; n = 16; s = 1
KS = 4 × 1; r = 3; n = 16; s = 1
1st FC layerN = 256; action function: ReLU
2nd FC layerN = 64; action function: ReLU
3rd FC layerN = 1; action function: sigmoid