It has been common for a process industry enterprise to make its management and economic profits better by researching and realizing the tecnology of advanced and optimized control in the rencent time.In this paper, we analyse the algorithm and characteristics of RBF artificial neural network and make some improvements on it. In addition, the method in which the input variables are identified with PCA(Principal Component Analysis)is specially introduced. Based on the the knowledge of distillation column and the field data,we found eight Soft-Sensing models of quality targets of distillation column with improved RBF neural network. It is obvious that the models have little error compared with values from the chromatogram and other devices.After found the Soft-Sensing models, we realize the optimization control and show the optimization setting value of the key control circuit to the operator.Besides the theory about System Identification and how to optimize the PID parameters of controller are talked. |