摘要
针对有约束模型预测控制在线计算量大的应用局限性,特别是在一些快速机电系统中应用的困难,利用显式预测控制"离线设计"和"在线综合"的特性,结合可达有效分区搜索算法,实现计算效率的进一步提升,同时保证系统实际运行的可靠性。针对强非线性、强耦合的欠驱动不稳定倒立摆系统,采用可达分区搜索与顺序搜索相结合的策略,建立了快速搜索显式预测控制律的综合实施方案。数值仿真和实时控制实验研究结果证实了该方法可在30 ms的采样周期内得到满足多种约束的计算结果,响应调节时间不超过2 s,并且适用范围提高25%。所得结果说明,基于可达分区与直接搜索的显式预测控制策略具有在快速机电系统上实施的可行性。
Extensive online computational complexity of constrained model predictive control （MPC） limits its practical application fields,especially in some fast electromechanical systems.Based upon the offline calculation and online synthesis characteristics of explicit MPC （EMPC）,the achievable region searching approach is employed to further speedup the online searching efficiency and raise the reliability.The achievable region searching and sequential searching methods are combined together to establish a comprehensive EMPC scheme to deal with a fast underactuated unstable inverted pendulum system with strong nonlinearity and coupling dynamics.Both mathematical simulations and real-time control experiments were performed to verify the effectiveness of the proposed scheme.The solu tion satisfying several constraints can be obtained within the sampling interval of 30 ms and the settling time is less than 2 s.Moreover,the applicable range is increased by 25%.These results suggest that the comprehensive EMPC scheme has application feasibility in fast electromechanical systems.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2014年第9期2037-2044,共8页
Chinese Journal of Scientific Instrument
基金
国家自然科学基金(61174094
61273138)
天津市自然科学基金(13JCYBJC17400)资助项目
关键词
显式预测控制
可达分区
计算复杂性
实时控制
倒立摆
explicit predictive control
achievable region
computational complexity
real-time control
inverted pendulum
作者简介
周昆,2006年于空军第一航空学院获得学士学位,现为95896部队助理工程师、南开大学硕士研究生.主要研究方向为模型预测控制及其应用.E-mail：blue_guard2014@126.com｜孙明玮（通讯作者）,2000年于南开大学获得博士学位,现为南开大学副教授,主要研究方向为自抗扰控制、模型预测控制、飞行器制导与控制和非线性优化.E-mail：smw_sunmingwei@163.com
