网络环境下随机Markov跳变系统的分布式预测控制研究

Networked stochastic Markovian jump linear system is a typically distributed hybrid system. This kind of system is complicated, and all subsystems are connected and coupled. In addition, the information transmission is always incomplete for the limitation of the network communication. Therefore, centralized model predictive control cannot meet the requirement of the networked uncertain distributed system. In order to solve this problem, distributed model predictive control is applied to study the networked stochastic Markov jump linear system. With the problems such as networked time delay, packet dropout and packet disordering taken into consideration, a stochastic switching system is modelled under certain switching sequences. Based on the modelled stochastic switching system, the distributed model predictive controller is designed for the local subsystem. Then, affects and restricts among the subsystems are coordinated, and the optimization of the whole distributed system is guaranteed by designing the performance index function under the expected significance and by choosing the weighted matrix. Finally, the theoretical results are verified by simulations, and are tried to be applied to the distributed tank model, which achieves the theory guiding practice. Based on random switching rule design, performance index function selection and optimization, the feasibility and optimality analysis, computational complexity and stability constraints and other aspects of the research, this project intends to promote theoretic research on distributed model predictive control algorithm and its application to the stochastic Markov jump systems.

网络环境下的随机Markov跳变系统是一类典型的分布式混杂系统，该系统结构复杂，子系统之间相互关联和耦合，并且网络通信能力受限会导致信息传输不完全，所以集中式预测控制方法不能很好地满足网络环境下不确定性分布式系统的要求。为了解决这一问题，本项目拟运用分布式预测控制方法，设计合适的切换规则，结合网络时延、数据丢包和数据包错序等问题发生的特点建立随机切换系统模型。在随机切换模型基础上，对子系统设计局部预测控制器，通过改进期望意义下的性能指标函数设计和加权矩阵的选取，协调好子系统之间的影响和制约，实现对整个分布式系统的优化调节。最后，将理论成果通过仿真验证并尝试应用于分布式水箱模型，达到理论指导实践。本项目欲通过随机切换规则的设计、性能指标函数的选取及优化、可行性与最优性分析、计算复杂度及闭环稳定性约束条件等几个方面的研究，推进分布式预测控制算法在随机Markov跳变系统中的应用。

网络环境下的随机Markov跳变系统是一类典型的分布式混杂系统。该系统结构复杂，子系统之间相互关联和耦合。此外，网络通信能力受限会导致信息传输不完全。因而，传统的集中式预测控制方法不能很好地满足网络环境下不确定性分布式系统的性能要求。为此，本项目以网络环境下随机Markov跳变系统为研究对象，根据混杂系统的特点，结合网络时延、数据丢包等网络诱导现象建立切换系统模型。然后，通过新的期望意义下的性能指标的提出和加权矩阵的选取，对子系统设计局部预测控制器。该分布式预测控制器的设计较好地协调了子系统之间的影响和制约，实现了对整个分布式系统的优化调节。最后，通过数值仿真和实验对理论结果进行了实验验证与优化，推进了分布式预测控制算法在网络化随机Markov跳变系统中的理论研究及其应用。