Subspace Method Aided Data-Driven Fault Detection Based on Principal Component Analysis

Table 4

TE process: fault detection for test data (%).

Fault	PCA		SMI-PCA			SMI-PCA KDE
		99.2	99.8	100	4.25	800	100	5.63	800
	98.0	98.6	99.46	3.88	800	99.5	5.25	800
	0	0	11.97	3.75		14.52	5.12
	4.4	96.2	100	4.62	800	100	6	800
	22.5	25.4	14.92	4.37		17.43	5.75
	98.9	100	100	4.62	800	100	6.12	800
	91.5	100	100	4.75	800	100	6.12	800
	96.6	97.6	97.70	3.62	811	97.81	5.12	811
	0	0	12.4	3.62		14.49	5.12
	33.4	34.1	81.56	3.62	811	82.82	5.12	811
	20.6	64.4	98.67	3.62	811	98.78	5.12	811
	97.1	97.5	43.35	3.62	811	46.76	5.12	811
	94.0	95.5	98.81	3.62	811	98.85	5.12	811
	84.2	100	99.82	3.62	801	99.86	5	801
	0	0	5.18	3.62		6.25	5.12
	16.6	24.5	4.46	3.62		5.9	5.12
	74.1	89.2	91.41	3.62	811	91.88	5.12	811
	88.7	89.9	30.63	3.62	811	32.49	5.12	811
	0.4	12.7	99.75	3.75	801	99.75	5.12	801
	26.4	43	89.25	3.62	811	90.08	5.12	811