摘要
提出一种基于观测器的故障检测滤波器的非凸多目标优化设计方法.针对线性时不变动态系统,构建一个由输出观测器和后滤波器组成的故障检测滤波器,将其残差动态特性描述为非凸的双线性矩阵不等式形式.利用双线性矩阵不等式中可完全平方非正定项的上界替代原非正定项,将双线性矩阵不等式转化为一组线性矩阵不等式,进而获得多目标优化问题的可解条件以及观测器增益与后滤波器矩阵的求解方法.仿真算例说明了算法的有效性.
A non-convex multi-objective optimal design approach to observer-based fault detection filters is proposed. A fault detection filter with an output observer and a post-filter is constructed for linear time invariant systems. The dynamics of generated residual is formulated as non-convex and in terms of bilinear matrix inequality (BMI). BMI terms can be translated into a serial of linear matrix inequalities (LMI) by applying the completing square into BMI to replace the non-positive terms by their upper bounds. The existence condition of multi-objective optimal problem and the solution of observer gain and post-filter matrix are also given and proved. The simulation results show the effectiveness of the proposed approach.
出处
《控制与决策》
EI
CSCD
北大核心
2005年第7期773-777,共5页
Control and Decision
基金
国家973计划项目(2002CB312200).