Experimental Investigation for Fault Diagnosis Based on a Hybrid Approach Using Wavelet Packet and Support Vector Classification

Table 4

Gear training samples.

Cases	Energy ratios in each frequency band
	1	2	3	4	5	6	7	8
Normal	0.5022	0.3977	0.0530	0.0077	0.0250	0.0098	0.0034	0.0012
	0.4511	0.4109	0.0716	0.0079	0.0407	0.0128	0.0038	0.0013
	0.4911	0.3799	0.0651	0.0080	0.0360	0.0150	0.0037	0.0011
Broken teeth	0.5591	0.3319	0.0615	0.0152	0.0224	0.0065	0.0029	0.0007
	0.6000	0.2979	0.0578	0.0130	0.0226	0.0060	0.0023	0.0005
	0.5589	0.3253	0.0686	0.0146	0.0216	0.0073	0.0031	0.0006
Wear	0.8194	0.1025	0.0226	0.0059	0.0266	0.0206	0.0018	0.0006
	0.8289	0.0902	0.0178	0.0056	0.0380	0.0173	0.0016	0.0005
	0.8080	0.1084	0.0209	0.0061	0.0391	0.0155	0.0018	0.0003
Missing teeth	0.2358	0.1971	0.2427	0.0528	0.1688	0.0814	0.0177	0.0035
	0.1239	0.2404	0.1706	0.0746	0.2052	0.1422	0.0387	0.0044
	0.2064	0.3353	0.1174	0.0537	0.1634	0.1051	0.0141	0.0046