Fault Diagnosis for Rotating Machinery Based on Convolutional Neural Network and Empirical Mode Decomposition

Table 4

Test accuracy of both classifiers on different features.

Features	SVM	Softmax
11 time-domain and EMD features	83.14%; (8647/10400)	82.93%; (8625/10400)
80 CNN features	99.05%; (10301/10400)	98.90%; (10286/10400)
91 combined features	99.75%; (10374/10400)	99.48%; (10346/10400)