Chemical process modeling based on support vector machine with flexible kernel function
the nonlinear model performance of support vector machine (SVM) is decided by the kernel function in all probability,and choosing a proper kernel function is very important.During the modeling process,usually by selecting several kernel functions,and optimizing its parameters,and then to select the one which gives the best performance of fitting and predicting.Unfortunately,this will lead to a very time-consuming procedure,and the result may not be optimal.Person Ⅶ function has fine flexibility,it can fit any smooth single-peak curve by adjusting its parameters,and it meets the Mercer conditions which is necessary for the valid kernel function.Therefore,Person VII flexible kernel function was introduced into SVM in order to avoid the selection of kernel function.The SVM based on Person VII flexible kernel function can simplifies the modeling process and improves the performance of the model.The improved SVM was applied to chemical process modeling,and satisfactory results were obtained.
support vector machine flexible kernel function chemical process modeling
Zheng Qifu Shen Xiaoli
College of Chemical and Material Engineering,Quzhou University,Quzhou 324000,China
国际会议
青岛
英文
235-238
2012-07-20（万方平台首次上网日期，不代表论文的发表时间）