聚酯工业生产缩聚过程的多目标优化  

Multi-objective optimal operation in polyester polymerization

作　　者：江沛[1] 曹柳林[1] 

机构地区：[1]北京化工大学信息科学与技术学院自动化系,北京100029

出　　处：《计算机与应用化学》2007年第11期1493-1496,共4页Computers and Applied Chemistry

基　　金：国家自然基金(60704011)

摘　　要：介绍一种聚酯(PET)生产缩聚过程的多目标优化技术,以最终产品的产量最大和质量最佳为优化目标,以第三和第四缩聚釜为优化对象,运用复合神经网络技术及相关机理知识建立优化模型。借鉴非劣分层方法、精英策略和群智思想,建立混合优化算法,通过该算法及惩罚函数寻优,设定生产过程中操作变量的最优参数,采用实际工业生产数据仿真和验证,建立周期性的两级优化结构来实现优化控制。A hybrid algorithm was developed for multi-objective optimization in chemical industry. It contained non-dominated sorting method, elitist strategy and swarm intelligence idea. The objectives in polyester polymerization were maximum yield and best quality： desired Intrinsic Viscosity and Molecular Weight Distribution. Models for optimization were founded by multi-neural networks. The hybrid algorithm has quick convergent rate and better diversity of solutions. By using this algorithm and industrial data, we get best solutions in polyester polymerization. We also put this optimization into practice by two-order optimal operation.

关 键 词：多目标优化 复合神经网络 非劣分层 粒子群 聚酯 

分 类 号：TQ316.47[化学工程—高聚物工业]