论文部分内容阅读
Some limitations exist in modeling of chemical process by SVM (support vector machine), such as unsatisfactory modeling accuracy, generalization ability and difficult kernel parameter selection process. To overcome these difficulties, combining wavelet SVM (W-SVM) with Morlet wavelet kernel was described in this article, reproducing kernel space and the wavelet analysis, and the proof to supporting Morlet wavelet kernel function was also given. To show the wavelet kernel function’s better generalization capability and accuracy, the proposed method was applied to establish a soft-sensor model for average molecular weight in polyacrylonitrile (PAN) production process. The results of real data simulation show that this method is effective