Soft Sensor Based on Kernel Principal Component Analysis and Radial Basis Function Neural Network for Microbiological Fermentation

To solve the problems of the high dimension and heavy correlation of inputs, a novel method of soft sensor based on the integration of both kernel principal component analysis (KPCA) and radial basis function neural network (RBFNN) is proposed and applied in the modeling of biomass estimation in the fermentation process of Avermectin in this paper. KPCA is a nonlinear PCA developed by using the kernel method, taking into account higher order information of the original inputs. The features extracted by KPCA fully show the complex relation between the original inputs of biomass estimation in the fermentation process of Avermectin. Industrial application showed that RBFNN by feature extraction using PCA and KPCA can perform better than that without feature extraction. Further more, better generalization performance in KPCA feature extraction than PCA feature extraction also showed that KPCA is a better solution of feature extraction in nonlinear data