Aiming at the fault diagnosis in a class of discrete linear time-varying system, a novel fault detection and
estimation algorithm is proposed. This algorithm uses an introduced virtual fault to construct a discrete fault tracking
estimator, and then uses the iterative learning algorithm to regulate the virtue fault close to the practical fault based on the
difference between outputs of the estimator and the practical system within the chosen optimization time, thereby reaching
the end of the fault diagnosis. This algorithm not only can detect the different-type fault of the system, but also can estimate
the fault signal accurately. Simulation results show the effectiveness of this algorithm.