Analytical Redundancy of Variable Cycle Engine Based on Proper Net considering Multiple Input Variables and the Whole Engine’s Degradation

Table 3

Performances of different convolutional neural networks.

Net	Average relative error/%	Layers	Training time/min (10 epochs)	Calculation time/s (1000 points)	Macroscopic structure
Proper net	0.841st	491st	251st	0.36451st
Dense net	1.17	707	535	1.4221
Mobile net	1.25	153	73	0.3716
Res101	1.33	143	155	0.9655
Alex net	12.71	24	6	0.1284
Vgg19 net	7.43	40	26	0.5133
Google net	2.05	143	40	0.2317
Squeeze net	3.85	68	17	0.1192

Yellow rectangle: convolution layer; red rectangle: activation layer; violet rectangle: pooling layer; green rectangle: normalization layer; white rectangle: fusion layer.