Iterative Learning Fault Estimation Algorithm for Time-delay Systems Based on Extended Observer

Abstract

For a class of multivariable linear, time-delay systems with actuator fault and measurement bounded disturbances in output, an iterative learning fault estimation algorithm based on extended observer is proposed. The extended observer is designed in terms of the linear matrix inequality technique such that the states and disturbances can be estimated simultaneously in every trials, then the faults and disturbances can be separated for avoiding impact to each other. Afterwards, the iterative learning fault estimation algorithm by defining estimation residual is chosen to adaptively approximate the actuator fault with initial error, then the necessary and sufficient conditions for the existence of the learning algorithm is given through λ norm theory and Bellman-Gronwall inequality, and the uniform convergence criteria of the control algorithm is also discussed. Simulation results verify the feasibility and effectiveness of this algorithm.