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

Table 7

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

F-measure over VALID with PSO-optimized 𝜀opt	Classification technique
	PNN		RVM		PCRVM
	𝜀opt	𝐹me	𝜀opt	𝐹me	𝜀opt	𝐹me
Feature extraction
None	0.6934	0.7711	0.6877	0.78014	0.6993	0.80563
DK	0.6983	0.77132	0.6919	0.80532	0.7154	0.83011
FFT	0.6916	0.78932	0.6997	0.81829	0.7278	0.82622
WPT + PCA; (Haar, level 𝐽=9)	0.7024	0.79451	0.7068	0.82788	0.7177	0.87112
DK + FFT	0.7161	0.78398	0.6949	0.83129	0.7269	0.84941
DK + WPT + PCA; (Haar, level 𝐽=9)	0.7116	0.82561	0.7114	0.85445	0.7147	0.88911

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