摘要
针对一阶最优性必要条件跟踪法优化间隔较长的缺点,提出一种基于输出反馈的显式实时优化方法.对系统进行不同工况下的离线优化,经函数拟合得到最优控制输入与输出变量的显式回归模型,直接应用于实时优化,避免了在线梯度估计.研究一个连续搅拌釜式反应器的反应过程,并对比两种方法的优化效果,结果验证了所提出方法的实际使用效果.
To circumvent the shortcoming of necessary conditions of optimality（NCO） tracking, namely that the optimizing period is relatively long, an explicit optimizing control approach based on output feedback is proposed. A number of off-line optimizations are carded out under various operating conditions, and the explicit model between optimal input and output variables is numerically established via function fitting. The model is used for online implementation and thus online gradient estimation is avoided. A continuous stirred tank reactor（CSTR） case is studied to demonstrate the advantages of proposed approach.
出处
《控制与决策》
EI
CSCD
北大核心
2013年第11期1697-1701,共5页
Control and Decision
基金
国家自然科学基金项目(61304081)
国家973计划项目(2012CB720505)
浙江省自然科学基金项目(LQ13F030007)
宁波市创新团队项目(2012B82002)
关键词
实时优化
输出反馈
一阶最优性必要条件跟踪
优化控制
real-time optimization
output feedback
necessary conditions of optimality tracking
optimizing control
