Research on soft sensor model based on Kernel Function Principal Component Analysis For gas outburst

Because gas accidents happen usually, gas outburst prediction becomes first issue to be solved urgently. In this paper soft sensor model has established based on KPCA (kernel function principal component analysis) and RBF NN. Through a nonlinear mapping function, the data was projected from the input space to feature space, then feature information of input variables are extracted. KPCA can deal with nonlinear data effectively. Compared with soft sensor model based on PCA (principal component analysis) and RBF NN, it has upper precision and generalization performance. Field application proves soft sensor model based on KPCA and RBF NN is effective and superior to traditional model.