针对一类线性不确定系统的故障诊断问题,提出了一种新的故障检测与估计算法.采用阈值限定技术,在选取的优化时域内,利用残差信号以及迭代轴上相邻两次残差的差分信号,对引入的虚拟故障信号进行逐次修正,使虚拟故障逼近系统中实际发生的故障,从而达到对系统故障估计的目的,并给出了算法在λ范数度量意义下收敛的充分条件.该算法不仅实现了不确定系统的故障检测与估计,还避免了因使用微分运算给系统带来的不良影响,取得了比单纯比例型迭代学习算法更快的收敛速度. Aiming at the problem of fault diagnosis for a class of linear uncertain systems, a new fault detection and estimation algorithm is proposed. By using the threshold definition technique, the residual signal and the adjacent two residuals Of the differential signal, the introduced virtual fault signal is modified one by one, the virtual fault approximation of the actual fault in the system, so as to achieve the purpose of system fault estimation, and gives the sufficient conditions for the algorithm to converge in the sense of λ norm The algorithm not only realizes the fault detection and estimation of uncertain system, but also avoids the adverse effect brought by the use of differential operation to the system, and achieves faster convergence rate than the simple proportional iterative learning algorithm.