论文部分内容阅读
为了提高间歇过程批次之间产品的一致性,并及时发现过程中的异常情况,提出一种基于过程数据相似度的多变量统计监控方法对间歇过程的操作进行在线监控。该方法将正常批次轨迹与参考批次轨迹之间的相似度作为一种新的监控指标,并利用核密度方法估计相似度的概率密度函数,计算出控制限,在批次反应过程中利用Kalman滤波器对当前批次的数据进行实时的估计从而实现在线监控。该方法和传统多向主元分析方法的监控性能在一个青霉素发酵仿真系统上进行了比较。仿真结果表明:该方法检测出渐变型扰动比MPCA方法提前了30h。
In order to improve the consistency of products between batches of batch process and find out the anomalies in the process, a multivariate statistical monitoring method based on process data similarity is proposed to monitor the operation of batch processes online. In this method, the similarity between the normal batch track and the reference batch track is taken as a new monitoring index, and the kernel density method is used to estimate the probability density function of similarity. The control limit is calculated and utilized in the batch reaction process Kalman filter real-time estimation of the current batch of data in order to achieve online monitoring. The monitoring performance of this method and the traditional multi-directional principal component analysis method are compared on a penicillin fermentation simulation system. The simulation results show that the proposed method can detect gradual perturbations earlier than the MPCA method by 30h.