Modeling of neutralizing process based on pruning fuzzy Least Squares Support Vector Machine

A pruning fuzzy Least Squares Support Vector Machine (LS-SVM) is proposed and applied into regression modeling in this paper. In the proposed algorithm, the squares of errors are weighted in objective function and each weight is associated with the corresponding absolute value of Lagrange multiplier. Otherwise, an iterative pruning approach which gradually removing some less important points is used in the process to gain spareness approximation. Hence, the pruning fuzzy LS-SVM can both satisfy the requirement of precision and spareness. The experiment result of a pH neutralizing process indicates the practicability and effectiveness of the presented algorithm.