摘要
针对基于纳什最优的分布式模型预测控制求解算法中存在的迭代次数多、收敛精度不高的缺点,提出了一种基于非合作博弈的分布式模型预测控制优化算法。该方法借鉴非合作博弈论中的针锋相对策略,将每个子系统看作博弈的参与者,在线优化过程中,各个子系统在该策略影响下使所有参与者更快促成合作,从而快速求得整体最优解。仿真表明,与传统的基于纳什最优的迭代求解相比,在给定精度情况下,提出的算法所需的迭代次数要低于传统算法;在给定迭代次数情况下,提出的算法的跟踪性能更优,在外界产生随机扰动时,该算法也具有较好的抗干扰能力。此外,将提出的算法应用于设施环境控制系统中,进一步说明了算法的有效性。
In order to overcome the shortcomings on Nash optimality has a low convergence speed and tion algorithm for distributed model predictive contro that the distributed model predictive control based ow convergence accuracy, we propose an optimizabased on non-cooperative game. The algorithm refers to the tit-for-tat strategy in the non-cooperative game theory, and each subsystem can be viewed as a participant of the game. During the process of optimization, each subsystem makes all participants quickly to cooperate under the tit-for-tat strategy, thus obtaining the overall optimal solution. Simulation results show that compared with the traditional method based on Nash optimality, the proposed algorithm can reduce iterations on average when a precision is given; and it has a better tracking performance when the number of iterations is given. In addition, when external disturbances occur randomly, this algorithm also has a good disturbance rejection capability. The application of the proposed algorithm in the facility environmental control system further proves its validity.
出处
《计算机工程与科学》
CSCD
北大核心
2016年第7期1484-1494,共11页
Computer Engineering & Science
基金
中央高校基本科研业务费专项资金(KYZ201421)
江苏省农业"三新"工程项目(SXGC[2014]309)
国家科技支撑计划-国家重点监管产品(乳制品
肉制品
白酒)电子溯源技术应用研究与示范
关键词
分布式模型预测控制
非合作博弈
针锋相对策略
纳什最优
distributed model predictive control
non-cooperative game
tit for tat strategy
Nash optimality