东北大学 信息科学与工程学院,辽宁 沈阳 110819
收稿日期:2015-12-08
基金项目:国家自然科学基金资助项目 (61573088,61433004)。
作者简介:齐文海 (1986-), 男, 山东泰安人, 东北大学博士研究生;
高宪文 (1954-), 男, 辽宁盘锦人, 东北大学教授, 博士生导师。
摘要:研究一类带有执行器饱和的Markov跳变系统的镇定问题, 转移概率是分段齐次的.首先, 通过建立合适的Lyapunov泛函, 运用椭球不变集估计系统均方意义的吸引域, 得到由线性矩阵不等式约束的闭环系统随机稳定的充分条件.然后, 通过求解凸优化问题得到状态反馈控制器增益及均方意义下吸引域的最大估计值.最后, 数值算例验证了所得结论的有效性.
关键词:执行器饱和Markov跳变系统分段齐次线性矩阵不等式凸优化
Stabilization for Piecewise Homogeneous Markov Jump Systems Subject to Actuator Saturation
QI Wen-hai, LI Xin, GAO Xian-wen
School of Information Science & Engineering, Northeastern University, Shenyang 110819
Corresponding author: QI Wen-hai, E-mail: qiwhtanedu@163.com
Abstract: The stabilization problem was studied for a class of Markov jump linear systems subject to actuator saturation, whose transition rates are piecewise homogeneous. Firstly, by using appropriate Lyapunov functional and ellipsoidal invariant set theory, the attraction domain of system in mean square sense was estimated to get the sufficient conditions with constraints of linear matrix inequalities for the closed-loop systems. Then, a convex optimization problem was solved to get the maximum domain of attraction in mean square sense and the state feedback controller gain. Finally, the effectiveness of the results was verified by a numerical example.
Key Words: actuator saturationMarkov jump systemspiecewise homogeneouslinear matrix inequalitiesconvex optimization
由于经常受到随机突变诸如外界随机干扰、内部元件的随机故障和失效等影响, 实际系统可以用Markov跳变系统来刻画.它是一类包含多个模态的重要随机混杂系统, 有着广泛的应用,例如网络控制系统[1]、机械系统[2]、故障检测系统[3]和经济系统[4]等.
转移概率 (TPs) 作为Markov跳变系统的一个关键性因素,直接影响系统性能.若Markov跳变系统的转移概率矩阵不随时间t发生变化, 即转移概率与t是相互独立的, 则为齐次Markov过程, 除此之外则被称为非齐次Markov过程[5].近20年来, 针对齐次Markov跳变系统取得了很多研究成果[6-9], 它们均假定Markov跳变过程满足齐次性, 然而转移概率在实际系统运行过程中很难长时间保持恒定.例如系统工程中的组件故障率、网络控制系统的随机丢包和时延等问题, 此类系统子模态之间的切换规律符合分段齐次Markov过程, 它是非齐次Markov跳变过程的一种特殊情况, 意味着转移概率随时间变化但在一定时间间隔内保持不变.由于考虑分段齐次转移概率能更好地描述许多实际系统的特性, 近几年, 分段齐次Markov跳变系统的研究逐渐成为热点[5, 10-12].
另一方面, 执行器饱和的存在严重影响系统性能甚至导致系统不稳定, 例如平衡指针[13-14]、小车弹簧摆系统[13, 15]、F-8飞行器[13, 16]、RLC电路[17]等.近几年, 越来越多的学者研究具有执行器饱和的Markov跳变系统, 取得丰硕的成果[18-21].然而, 却没有关于具有执行器饱和的分段齐次Markov跳变系统的文献报道.
1 问题描述及相关引理考虑一类Markov饱和跳变系统:
(1) |
考虑δt, 意味着转移概率是时变的.同时, 假设δt为t的分段常函数.跳变转移概率矩阵定义为
注1 Markov跳变过程的分段齐次转移概率矩阵Λ(δt+h)是时变转移概率矩阵的一种特殊情况, 转移概率随时间变化但在一定时间间隔内保持不变.
类似地, 参数{δt, t≥0}也是一个Markov跳变过程, 随t在有限集合Γ={1, 2, …, M}中取值.
(3) |
函数σ(·):Rm→Rm是标准的向量饱和函数, 即
对于任意的rt=i和δt=k, 为了简化记号, A(rt), B(rt) 记为Ai, Bi.
设计参数依赖的状态反馈控制器为
(4) |
定义1[19]对任意的初始模态rt∈Γ, 初始状态x0∈Ψ, Ψ?Rn下, 使得
对于任意矩阵Pi, 定义椭圆
(5) |
考虑控制器 (4), 可以得到闭环系统:
(6) |
(7) |
(8) |
证明
2.2 状态控制器的设计和吸引域估计本节采用椭圆不变集来估计系统的吸引域, 在吸引域中求解最大的作为系统的吸引域估计值.令参考集χR?Rn为一个包含原点的凸集.对于包含原点的集合φ?Rn, 定义
定理1给出了系统 (6) 随机稳定的充分条件, 需要将这些充分条件转化为便于求解的线性矩阵不等式的形式, 进而求得状态反馈控制增益Fi, k和最大不变吸引域.另外, 通过求解下列凸优化问题,验证给定的初始状态x0∈Rn是否在C0{x1, x2, …, xω}内.
(9) |
(10) |
(11) |
对于i∈S1,k∈Γ,由于存在设计参数Fi, k, Hi, k, 不等式 (7) 是非线性的,对不等式左边分别左乘和右乘Qi, k得
(12) |
(13) |
(14) |
最后优化问题 (8) 转化为如下线性矩阵不等式形式的优化问题:
(15) |
3 数值仿真用一个数值算例来验证主要结论的有效性.假设执行器饱和的分段齐次Markov跳变系统具有两个模态,即S={1, 2}, 其参数矩阵为
图 1(Fig. 1)
图 1 系统模态Fig.1 System mode |
图 2(Fig. 2)
图 2 上层切换Fig.2 High-level switching |
图 3(Fig. 3)
图 3 状态轨迹Fig.3 State trajectories |
注2通过求解优化问题 (14), 可以验证初始状态满足吸引域条件.通过转移概率矩阵Π, 由Matlab仿真可以得到图 2.当图 2中的纵坐标为1时,考虑Λ1对系统的影响;当纵坐标为2, 考虑Λ2对系统的影响.以此类推得到图 1.转移概率矩阵Π可以作为上层随机切换, 控制下层Λ1,Λ2,Λ3,Λ4之间的切换.
4 结论针对具有执行器饱和的Markov跳变系统, 在考虑分段齐次转移概率的情况下, 构造系统均方意义下的稳定域, 在线性矩阵不等式的框架下, 实现了控制器增益和吸引域最大估计值的求解.数值仿真进一步验证了所得结论的有效性.
参考文献
[1] | Dong J W, Kim W J. Markov-chain-based output feedback control for stabilization of networked control systems with random time delays and packet losses[J].International Journal of Control, Automation and Systems, 2012, 10(5): 1013–1022.DOI:10.1007/s12555-012-0519-x |
[2] | Shen L J, Buscher U. Solving the serial batching problem in job shop manufacturing systems[J].European Journal of Operational Research, 2012, 221(1): 14–26.DOI:10.1016/j.ejor.2012.03.001 |
[3] | Yin Y, Shi P, Liu F, et al. Gain-scheduled fault detection on stochastic nonlinear systems with partially known transition jump rates[J].Nonlinear Analysis:Real World Applications, 2012, 13: 359–369.DOI:10.1016/j.nonrwa.2011.07.043 |
[4] | Mao X R, Yuan C G. Stochastic differential equations with Markovian switching[M]. London: Imperial College Press, 2006. |
[5] | Zhang L X. H∞ estimation for discrete-time piecewise homogeneous Markov jump linear systems[J].Automation, 2009, 45(11): 2570–2576.DOI:10.1016/j.automatica.2009.07.004 |
[6] | Mao X R. Stability of stochastic differential equations with Markovian switching[J].Stochastic Process and Their Applications, 1999, 79(1): 45–67.DOI:10.1016/S0304-4149(98)00070-2 |
[7] | Boukas E K. Stochastic switching systems:analysis and design[M]. Berlin: Birkhauser, 2005. |
[8] | Zhang L, Boukas E K. Stability and stabilization of Markovian jump linear systems with partly unknown transition probabilities[J].Automatica, 2009, 45(2): 463–468.DOI:10.1016/j.automatica.2008.08.010 |
[9] | Zhang Y, He Y, Wu M, et al. Stabilization for Markovian jump systems with partial information on transition probability based on free-connection weighting matrices[J].Automatica, 2011, 47(1): 79–84.DOI:10.1016/j.automatica.2010.09.009 |
[10] | Ding Y C, Liu H. Filtering-based fault detection for a class of piecewise homogeneous Markov jump systems[J].Mechatronics and Control, 2014: 759–764. |
[11] | Zhang L.H-infinity control of a class of piecewise homogeneous Markov jump linear systems[C]// Proceedings of the 7th Asian Control Conference.Hong Kong, 2009:197-202. |
[12] | Wu Z G, Ju H P, Su H, et al. Stochastic stability analysis of piecewise homogeneous Markovian jump neural networks with mixed time-delays[J].Journal of the Franklin Institute, 2012, 349(6): 2136–2150.DOI:10.1016/j.jfranklin.2012.03.005 |
[13] | Tarbouriech S, Garcia G. Stability and stabilization of linear systems with saturating actuators[M]. London: Springer-Verlag, 2011. |
[14] | Khalil H K. Nonlinear systems[M]. London: Macmillan, 1992. |
[15] | Grimm G, Hatfield J, Postlethwaite I, et al. Anti-windup for stable linear systems with input saturation:an LMI based synthesis[J].IEEE Transactions on Automatic Control, 2003, 48(9): 1509–1525.DOI:10.1109/TAC.2003.816965 |
[16] | Wu F, Grigoriadis K M, Packard A. Anti-windup controller design using linear parameter-varying control methods[J].International Journal of Control, 2000, 73(12): 1104–1114.DOI:10.1080/002071700414211 |
[17] | Zhao J J, Wang J, Park J H, et al. Memory feedback controller design for stochastic Markov jump distributed delay systems with input saturation and partially known transition rates[J].Nonlinear Analysis:Hybrid Systems, 2015, 15: 52–62.DOI:10.1016/j.nahs.2014.07.002 |
[18] | Boukas E K. Stochastic output feedback of uncertain time-delay systems with saturating actuators[J].Journal of Optimization Theory and Applications, 2003, 118(2): 255–273.DOI:10.1023/A:1025443220763 |
[19] | Liu H P, Boukas E K, Sun F C, et al. Controller design for Markov jumping systems subject to actuator saturation[J].Automatica, 2006, 42(3): 459–465.DOI:10.1016/j.automatica.2005.10.017 |
[20] | Li J N, Pan Y J, Su H Y, et al. Stochastic reliable control of a class of networked control systems with actuator faults and input saturation[J].International Journal of Control, Automation and Systems, 2014, 12(3): 564–571.DOI:10.1007/s12555-013-0371-7 |
[21] | Wang Y J, Zuo Z Q, Cui Y L.Stochastic stabilization of Markovian jump systems in the presence of partial unknown transition probabilities and actuator saturation[C]// Proceedings of 29th Chinese Control Conference.Beijing, 2010:29-31. |