间歇过程滑动窗口子时段PCA建模和在线监测

Sub-stage PCA Modeling and On-line Monitoring with Moving Windows for Batch Processes

针对在短期内不容易获得充足建模数据的间歇工业过程,提出一种间歇过程监视方法,该方法只需要一次正常间歇操作数据,利用滑动窗口,进行子时段划分、建立初始主元分析(PCA)监测模型,同时提出了基于子时段PCA模型的在线监测算法和一种模型更新策略.通过在注塑过程的成功应用,表明了所提出方法的可行性和有效性.

For batch processes with difficulties in collecting sufficient data in limited time,a batch monitoring method is proposed. The proposed method uses a single normal batch reference data and moving windows to divide sub-stage and builds an initial PCA monitoring model, The on-line monitoring method and method updating the model are proposed. The application to an injection molding process shows the effectiveness and feasibility of the proposed method for batch industry.