摘要
研究一种批次过程阶段划分方法.首先,将间歇过程数据按照批次方向展开成二维数据,某采样时刻到末采样时刻的数据构成对应于该时刻的数据段.然后对每一数据段进行主元分析,获得第一主元贡献率.最后,采用第一主元贡献率的变化来表征批次内部阶段的转变,对批次过程划分阶段.将该方法应用于青霉素仿真发酵过程,分段结果明显与实际菌体生长阶段一致.将该方法应用于实际的半导体生产过程数据中,分段结果验证了分段理论的有效性.
In this article, a method of dividing batches into sub-phases was researched. First, the batches of data were unfolded into two-dimention based on the batch direction. AU the data from the certain sample time to the last formed the data phase correspond to the sample time. The cumulative percent variance of the first principle component) could be obtained after principal component analysis was applied to all the data phases. Then the value change of the was used as an indicator of phase changing. This method was applied to the data collected from penicillin simulation batch process. The data was divided into phases abviously and showed consistent with the thalli' s growth. In the end, the proposed method was applied to the actual semiconductor production process data as well, and the segment results verified the effectiveness of the algorithm.
出处
《沈阳化工大学学报》
CAS
2013年第1期63-68,共6页
Journal of Shenyang University of Chemical Technology
基金
国家自然科学基金资助项目(61034006)