基于滑动时间窗口加权MPCA的间歇过程监测方法

针对间歇过程的时段划分问题,提出了一种滑动时间窗加权MPCA方法,利用了相邻观测数据所蕴含过程信息的相关性,降低了过程扰动、观测噪声以及多阶段间过渡过程等不确定性因素对时段划分的影响,进而改善了间歇过程的监测效果。将所提出的方法应用于青霉素发酵过程阶段划分和监测,实验结果表明,所提出方法与常规MPCA相比,能够得到更好的效果。

A method of batch process monitoring based on Sliding-time-window weighted MPCA

Using the correlation of adjacent data contained in the formation process, we propose a new method involving a sliding-time-window weighted MPCA (SWMPCA) to reduce the effect of the process noise, transitions in the multi-stage process, and other uncertainty factors in the stage of division in batch processes. As a result, the monitoring performance of a batch process is improved. The proposed method of improved MPCA has been employed in the division stage of the penicillin fermentation process and monitoring experiments. The results show that the proposed SWMPCA method is better than conventional MPCA method for batch process monitoring.