基于搜索法的GPC控制器参数的鲁棒优化

New Method of Robust Optimization of Generalized Predictive Controller Parameters by Searching Algorithm

为了对广义预测控制器(GPC)的两个具有连续性的参数,柔化因子和控制加权因子,进行优化设计,从GPC闭环反馈结构出发,由小增益定理推导出一个GPC鲁棒稳定性的指标。以此为主,辅以快速性要求,构建了一个优化控制的目标函数,然后采用单纯形搜索法,对柔化因子和控制加权进行优化设计。仿真结果证明了此方法的有效性和对各种模型的建模误差的适应性,但目前不适用于不稳定对象的GPC参数优化,有待改进。

The aim of the research is to optimize the two contirmative parameters of GPC control systems, softening factor and control weighting factor. Based on the closed-loop feedback structure an index of robust stability of GPC control systems is formulated according to the small-gain criterion. By multiplying a factor indicating the swiftness of system response, that index is changed into a cost function to optimize the two parameters . The simplex algorithm is used in the optimization process. And simulation work also illustrates this method's feasibility and compatibility with several patterns of modeling error. It has to be mentioned that this method is limited to GPC control systems with stable plants, and need to be extended to unstable plant cases.