论文部分内容阅读
The modeling and monitoring methods of nonlinear multimode transition process are proposed. The traditional methods such as principle component analysis (PCA) and partial least squares (PLS) are not effective for multimode problem. In order to solve this multimode problem, a new reconstruction-based approach is proposed for modeling and monitoring purpose. The characteristics of the multimode transition process are analyzed in this paper. In different batch cycles transition patterns may follow different trajecto-ries and reveal different characteristics. After the operation patterns change, the two neighboring modes still share some common cor-relations. According to the principle, the reconstruction-based algorithm is proposed. The modeling and monitoring of multimode tran-sition process are proposed based on the reconstruction algorithm. The simulation results show the proposed methods are effective.