Simultaneous-Fault Diagnosis of Automotive Engine Ignition Systems Using Prior Domain Knowledge and Relevance Vector Machine

Table 6

Evaluation of different combinations of techniques using GA-optimized threshold.

F-measure over VALID with GA-optimized 𝜀opt	Classification technique
	PNN		RVM		PCRVM
	𝜀opt	𝐹me	𝜀opt	𝐹me	𝜀opt	𝐹me
Feature Extraction
None	0.6875	0.76114	0.6757	0.79604	0.7005	0.81244
DK	0.7193	0.79819	0.6813	0.80325	0.7167	0.83023
FFT	0.6791	0.79236	0.7401	0.82129	0.7353	0.82621
WPT + PCA; (Haar, level 𝐽=9)	0.6998	0.79125	0.7161	0.83118	0.7234	0.85816
DK + FFT	0.7245	0.79123	0.6755	0.82212	0.7291	0.85414
DK + WPT + PCA; (Haar, level 𝐽=9)	0.7313	0.81431	0.7023	0.84524	0.7124	0.87113

*The experiment was run for 20 times, and the best 𝜀_opt and 𝐹_me were returned.