μ分析的线性矩阵不等式方法及其优化算法  

Optimized LMI algorithm for μ analysis

作　　者：富饶[1] 杨莹[1] 黄琳[1] 

机构地区：[1]北京大学力学与工程科学系,北京100871

出　　处：《控制与决策》2004年第3期247-251,共5页Control and Decision

基　　金：国家自然科学基金资助项目(10272001);国家重点基础研究专项经费资助项目(G1998020302).

摘　　要：通过采用S-过程和投影引理,得到了结构奇异值上界的LMI判据.该判据是基于状态空间描述的,从而消除了频率扫描过程和频率响应曲线拟合过程,并具有较好的数值性态.以该判据为基础,给出了计算结构奇异值上界的优化投影迭代算法,并将该方法应用于基准测试系统和典型电力系统,以验证其有效性.数值结果表明,该方法与经典频域方法和状态空间方法相比具有更好的求解效率.By using S-procedure and projection lemma, new μ upper bounds for a dynamic system are derived in terms of LMIs. Since the approach is based on state-space, it requires no frequency sweep and frequency response curve fitting and therefore has a better numerical property. A special optimal projective iteration algorithm for computing the μ upper bound is derived. To demonstrate the effectiveness, this approach is applied to benchmark system and typical power system. Numerical simulation results of this approach are compared with those of classical frequency domain methods and state-methods.

关 键 词：结构奇异值 线性矩阵不等式 半正定规划 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]