Soft sensor technique using LS-SVM and standard SVM

Support vector machine (SVM) is a modern machine learning method based on Vapnik´s statistical learning theory. In this paper, regression support vector machine has been proposed as a tool to soft sensor technique, in which SVM is used to estimate variable which is highly nonlinear. An introduction to standard SVM and LS-SVM is given at first, then uses them to identify absorption stabilization system (ASS) process variable. Systematic analysis and case studies are performed and indicate that the proposed method provides satisfactory performance with excellent approximation and generalization property, soft sensor technique based on SVM achieves superior performance to the conventional method based on neural networks.