A Framework for Diagnosing the Out-of-Control Signals in Multivariate Process Using Optimized Support Vector Machines

Table 1

Correct classification percentages (%) of the optimization SVM, the SVM, and the BPNN.

Model	Shift	Correct classification percentage (%)
	magnitude	(0 0 0 0)	(0 1 0 0)	(0 0 1 0)	(0 0 0 1)	Percentage
BPNN	(0.0 0.0)	470	14	13	3	94.00
	(1.0 0.0)	0	450	12	38	90.00
	(0.0 1.0)	0	3	452	45	90.38
	(1.0 1.0)	0	30	25	445	89.00
	(1.5 0.0)	0	460	1	39	92.00
	(0.0 1.5)	0	2	458	40	91.60
	(1.5 1.5)	0	27	23	450	90.00
	(2.0 0.0)	0	468	0	32	93.6
	(0.0 2.0)	0	5	465	30	93.0
	(2.0 2.0)	0	22	28	450	90.0
	(2.5 0.0)	0	470	0	25	94.0
	(0.0 2.5)	0	3	472	25	94.4
	(2.5 2.5)	0	20	22	458	91.6
	(3.0 0.0)	0	475	2	10	94.0
	(0.0 3.0)	0	3	471	12	94.2
	(3.0 3.0)	0	25	15	460	92.0
Aggregate						92.11
SVM	(0.0,0.0)	480	8	6	6	96.0
	(1.0 0.0)	0	468	12	20	93.6
	(0.0 1.0)	0	9	468	23	93.6
	(1.0 1.0)	0	16	22	462	92.4
	(1.5 0.0)	0	461	6	33	92.2
	(0.0 1.5)	0	3	472	25	94.2
	(1.5 1.5)	0	12	13	475	95.0
	(2.0 0.0)	0	474	8	18	94.8
	(0.0 2.0)	0	3	470	27	94.0
	(2.0 2.0)	0	10	22	468	93.6
	(2.5 0.0)	0	480	2	18	96.0
	(0.0 2.5)	0	9	472	19	94.4
	(2.5 2.5)	0	22	18	460	92.0
	(3.0 0.0)	0	480	2	18	96.0
	(0.0 3.0)	0	2	478	20	95.6
	(3.0 3.0)	0	15	13	472	94.4
Aggregate						94.23
optimized	(0.0,0.0)	500	0	0	0	100
SVM	(1.0 0.0)	0	480	1	19	96.0
	(0.0 1.0)	0	3	484	13	96.8
	(1.0 1.0)	0	8	10	482	96.4
	(1.5 0.0)	0	491	0	9	98.2
	(0.0 1.5)	0	0	488	12	97.6
	(1.5 1.5)	0	5	5	490	98.0
	(2.0 0.0)	0	490	1	9	98.0
	(0.0 2.0)	0	1	488	11	97.6
	(2.0 2.0)	0	7	8	485	97.0
	(2.5 0.0)	0	490	0	10	98.0
	(0.0 2.5)	0	0	492	8	99.4
	(2.5 2.5)	0	5	4	491	99.2
	(3.0 0.0)	0	494	0	6	98.8
	(0.0 3.0)	0	0	495	5	99.0
	(3.0 3.0)	0	7	6	487	97.4
Aggregate						97.96