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

Table 9

Simultaneous fault diagnosis under 𝐹-measure with classifiers trained with single faults (with DK + WPT + PCA and 𝜀_opt = 0.7147).

Test dataset 𝑇𝐸𝑆𝑇𝑠; (simultaneous faults only)	Classification technique	Fault 2	Fault 3	Fault 5	Fault 6	Fault 7	Fault 8	Fault 9	Fault 10
Faults (2, 6)	PNN	0.6917	—	—	0.6848	—	—	—	—
	RVM	0.7128	—	—	0.7067	—	—	—	—
	PCRVM	0.7435	—	—	0.7392	—	—	—	—
Faults (5, 7)	PNN	—	—	0.7321	—	0.7054	—	—	—
	RVM	—	—	0.7522	—	0.7214	—	—	—
	PCRVM	—	—	0.7926	—	0.7557	—	—	—
Faults (3, 10)	PNN	—	0.8003	—	—	—	—	—	0.4854
	RVM	—	0.7943	—	—	—	—	—	0.6159
	PCRVM	—	0.7844	—	—	—	—	—	0.6995
Faults (2, 5, 9)	PNN	0.7434	—	0.7457	—	—	—	0.6809	—
	RVM	0.7123	—	0.7359	—	—	—	0.7256	—
	PCRVM	0.7847	—	0.7858	—	—	—	0.7704	—
Faults (3, 8, 10)	PNN	—	0.6842	—	—	—	0.7118	—	0.6803
	RVM	—	0.6724	—	—	—	0.7319	—	0.6931
	PCRVM	—	0.7349	—	—	—	0.7828	—	0.6993

—: Indicates that the probability is lower than 𝜀_opt and hence ignored.