论文部分内容阅读
针对间歇过程时段的切换存在过渡区域,同时,间歇过程数据有着强非线性的特点,提出一种基于时段及过渡区域的KICA间歇过程监测方法。该方法基于MPCA及k-means聚类算法对间歇过程进行子时段划分,并基于第一主元贡献率差值识别时段间的过渡区域,在此基础上,对稳定时段建立统一KICA监测模型,而过渡区域针对各时刻滑动窗口进行KICA建模监测。将该方法应用于青霉素发酵过程在线监测,实验结果表明,相比sub..PCA监测方法,本文基于时段及过渡区域的KICA监测方法能更及时、准确的检测到过渡区域的异常。
There is a transition region for the switching of the intermittent process, meanwhile, the intermittent process has a strong non-linear characteristic. A KICA intermittent process monitoring method based on time zone and transitional region is proposed. Based on the MPCA and k-means clustering algorithm, the method divides the batch process into sub-periods and identifies the transitional area based on the difference of the contribution rate of the first principal component. On this basis, a unified KICA monitoring model is established for the stable period, The transition area for each moment sliding window KICA modeling monitoring. The method was applied to on-line monitoring of penicillin fermentation process. The experimental results show that compared with sub-PCA monitoring method, the KICA monitoring method based on period and transition region can detect the transition region anomalies more timely and accurately.