双层结构模型预测控制算法性能分析

Performance analysis of double-layered model predictive control algorithm

双层结构模型预测控制是在常规模型预测控制的基础上集成了稳态目标计算层,所增加稳态目标计算层通过对给定外部目标进行调节或执行独立经济优化为底层动态控制提供设定值.本文着重于分析增加稳态目标计算层对常规模型预测控制策略在控制性能、经济性能、及鲁棒性方面的影响.阐明稳态目标计算是在动态不确定性下的最优决策过程,可实时考虑当前工况环境给出最佳设定值,使动态层控制量的计算更加合理,并可保证在动态控制层实现无静差控制;稳态目标计算层可利用包含目标跟踪和经济优化的性能指标,在保证跟踪控制要求的同时利用操作自由度进行经济性优化,有助于提高系统经济性能,此外由于操作自由度增加使得系统对于特定方向的扰动和通道失配具有更强的鲁棒性.最后通过实验仿真验证了所做分析的正确性.

Compared with conventional model predictive control,the double-layered model predictive control adds steady-state target calculation layer which provides setting values to dynamic control layer by adjusting a given external target or implementing independent economic optimization.This paper focuses on the analysis of the effects on the control performance,economic performance and robustness of adding a steady-state target calculation layer on conventional model predictive control strategies.It is stated that steady-state target calculation is the optimal decision process under dynamic uncertainty and by considering the current working conditions to calculate the best setting value,which makes the control variables more reasonable and offset-free control can be insured in the dynamic control layer also.The steady-state target calculation layer can use performance indicators including target tracking and economic optimization to optimize the system economy by using freedom of operation while ensuring the tracking control requirements.In addition,the system is more robust to disturbance and channel mismatches in specific directions due to the increase of operation freedom.Finally,the correctness of the analysis is verified by experimental simulation.