基于迭代学习的离散切换系统故障估计

针对一类具有任意切换序列的离散切换系统的故障估计问题,提出了一种新的故障估计算法.该算法利用引入的虚拟故障信号构建出故障估计器,并利用残差信号通过迭代学习方法对引入的虚拟故障进行逐次修正.使虚拟故障随着迭代次数的增加逐渐逼近实际故障.利用压缩映射方法严格证明了算法在各个子区间上的收敛性,给出了算法的收敛条件.理论分析表明,所提算法能够在有限区间上精确估计出切换系统发生的不同类型故障.最后通过仿真实验进一步验证了所提算法的有效性.

Fault estimation for discrete switched system based on iterative learning

Aiming at the problem of fault estimation in a class of time-varying discrete switched system with arbitrary sequence, in this paper we propose a novel fault estimation algorithm. The algorithm uses the introduced virtual fault signal to construct fault estimator, and uses the residual signal to modify the introduced virtual fault step by step through using the iterative learning method and making the virtual fault gradually approach to the actual fault by increasing the iterative number. The convergence of the algorithm in each subinterval is strictly proven by the use of contraction mapping method, and the convergent condition of the algorithm is provided. Theoretical analyses indicate that the proposed algorithm can estimate different types of faults occurring in a switched system accurately in a finite interval. Finally, the validity of the algorithm is verified by simulations.
Keywords： discrete-time switched systems iterative learning fault estimation arbitrary switched rules