论文部分内容阅读
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 multi-objective optimization technology for polycondensation of polyester (PET) is introduced. The objective of the optimization is to maximize the yield of the final product and optimize the quality. The third and fourth polycondensation reactors are optimized. The composite neural network technology and the related Mechanism knowledge to establish optimization model. By referring to non-inferior stratification method, elite strategy and group-wise mentality, a hybrid optimization algorithm is established. By optimizing the algorithm and penalty function, the optimal parameters of the manipulated variables in the production process are set up. The simulation and verification of actual industrial production data are used to establish Periodic two-level optimization of the structure to achieve optimal control.