Table 5: Fault diagnosis results with different combinations of features.

Feature combinationFault diagnosis accuracy of the 10-fold cross validation method (%)

94.6295.1794.2994.4093.8194.1193.5495.1295.0193.98
94.4893.6193.0193.7794.1595.2593.9994.5894.6293.71
92.0493.6293.1193.2591.3794.5493.2792.3293.2993.41
93.0293.5792.1592.8992.4694.1993.3293.5592.5392.77
KPCA1~KPCA493.0893.7893.0793.5792.1793.0492.8092.4293.4893.36
LLE1~LLE494.1093.5994.5194.2293.9894.1093.9993.4493.9894.78
F1 ~F992.6791.8093.5492.4792.1193.6492.9793.6294.0193.00