摘要
本文基于模式识别之观点研究了顺丁橡胶聚合反应过程,直接在由输入参量构筑的空间中,根据最佳输入模式的聚类特征筛选关键控制量,按本征值分解法拟合聚合物ML值符合控制要求45±2的模式代表点的分布结构,从而在新的坐标表象中以超平面识别模型优化控制聚合过程.
Based on the pattern recognition point of view, this paper concerns with the butadiene rubber polymerization process. The important manipulated variables were selected according to the clustering nature of the optimal input patterns in the pattern space. By fitting the distribution structure of optimal input patterns which have ML value of polymer satisfying the control limit 45卤2, a hyperplane recognition model for ML value was built to optimize the polymerization process in the new coordinate representation system.
出处
《自动化学报》
EI
CSCD
北大核心
1989年第1期1-7,共7页
Acta Automatica Sinica
