Table 2.

Comprehensive analysis of soft sensor models based on SVM (for the list of abbreviations see Table A1).

Ref	Prediction model	Compared with	RMSE/MSE	MAXE	Best Performance
[57]	LS-SVM	RBF-NN	-	-	LS-SVM
[45]	LS-SVM	NN	-	-	LS-SVM
[49]	SVM	BP-NN	-	-	SVM
[50]	PSO-LS-SVM		0.7835	0.486	PSO-LS-SVM
		LS-SVM	0.1032	1.493
[51]	PSO-SVM	SVM	-	-	PSO-SVM
[52]	ABC-MLS-SVM	PID control	-	-	ABC-MLS-SVM
[54]	GRA-LS-SVM		2.518	2.641	GRA-LS-SVM
		RBF-NN	14.273	6.271
		LS-SVM	3.219	2.847
		GRA-RBFNN	4.162	3.594
[56]	IPSO-SVM	PSO	-	-	IPSO-SVM
[58]	MLS-SVM Inversion	LS-SVM	-	-	MLS-SVM Inversion

Ref - References, RMSE – Root Mean Square Error, MAXE –Maximum Absolute Error, MSE – Mean Square Error