基于递归神经网络模型预测控制的模型平稳切换  

Smooth model switching scheme based on recurrent neural network model predictive control

作　　者：杜福银[1] 徐扬[1] 陈树伟[1] 

机构地区：[1]西南交通大学智能控制开发中心,四川成都610031

出　　处：《计算机应用》2006年第6期1398-1400,共3页journal of Computer Applications

基　　金：国家自然科学基金资助项目(60474022)

摘　　要：不同生产条件下的控制系统可视多模型控制系统,但多模型控制在模型切换时会引起系统的瞬态响应。采用递归神经网络建立系统的多个模型,基于模型预测控制进行控制模型切换,克服了模型切换时引起的系统瞬态响应,实现系统的平稳切换。并通过仿真表明这种切换策略明显改善了模型切换过程的动态性能。The control system of differ product conditions could be regarded as a multi-model control system. But model switching will result in transient response of the plant controlled. A method to establish multiples models of system using recurrent neural network and complete switch process based on model predictive control was proposed. Which avoid transient response of switch process and achieve model switching smoothly. Simulation performances indicate that the model switching scheme improve dynamic quality of switch process greatly.

关 键 词：比例积分微分控制器 递归神经网络 模型预测控制 模型切换 平稳 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]