冲突多目标相容预测控制

两层结构的相容控制框架可以解决复杂冲突多目标控制问题,但其需要克服的困难是多目标直接优化、第一层所选区间目标的实现和在线实时优化控制。采用多目标遗传进化算法解决了直接优化多目标的工具问题。为能实现第一层中所选择的控制目标,第二层中设计了不同于第一层中形式的指标,并加入选择函数以反映实际系统的需要。结合预测控制思想的特点,提出了在线迭代遗传算法,解决了在线优化速度问题。以预测控制动态过程中误差、能耗两个目标为例对相容预测算法进行了说明。

Conflicting Multi-objective Compatible Predictive Control

Two layer compatible control framework is an effective method to solve the conflicting multi-objective control problem, however, which should overcome the difficulties---direct multi-objective optimization, be assuring of the realization of the multiple objectives selecting from the first layer and online optimization. Multi-objective genetic algorithm (MOGA) is adopted as the optimization tool. To assure the objectives realization from the first layer, a different optimization performance index was designed and a selection function was added in to reflect the control system requirement. Integrated with the predictive control theory, online iterative genetic algorithm (IGA) was proposed and solved the optimization"speed"problem. The predictive control dynamic process error and energy consumption were taken as the two objectives to illuminate the compatible predictive control algorithm (CPCA).