机构地区: 北京化工大学自动化研究所
出 处: 《北京化工大学学报(自然科学版)》 2009年第4期100-104,共5页
摘 要: 以连续搅拌反应釜(CSTR)为例,通过对一般化学习网络(ULN)分支上时间延迟的优化训练,提高了CSTR这类具有滞后环节的复杂非线性系统的建模精度。将一般化学习网络和模糊理论相结合,提出了一种改进的神经网络内模控制方法(改进的ULN-IMC)。仿真结果表明,改进的ULN-IMC可有效的提高CSTR的跟踪和抗扰动定值控制过程。 Through the training of time delays on the branches of a universal learning network (ULN) to the optimal values, the modeling precision for a continuous stirred tank reactor (CSTR)-a complicated nonlinear system with a large lag-was greatly improved. Furthermore, an improved ULN internal model control method (improved ULN-IMC) based on fuzzy control theory has been proposed. In simulations of a CSTR using this new method, the tracking and fixed set-point control when subjected to an external disturbance showed good performances.
关 键 词: 一般化学习网络 连续搅拌反应釜 时间延迟 神经网络内模控制 模糊理论
领 域: [自动化与计算机技术] [自动化与计算机技术]