Unknown Input Observer Design of Switched Systems Based on Iterative Learning

For a class of discrete switched systems with unknown input, an unknown input observer design method is proposed under the premise of changes along time axis but no changes along iteration axis. This method applies the iterative learning control thought to the design of unknown input observer, constructs the unknown input observer by introducing virtual input signal, and uses the error signal generated from the actual system output and the observer output to correct repetitively the virtual input, which gradually approximates the actual unknown input as the iterations increase. Moreover, the convergence of the observer is strictly proved based on contraction mapping theory, as well as the convergence condition is given. The theoretical analysis indicates that designed unknown input observer can accurately estimate the state and unknown input of the system simultaneously. Simulation example further verifies the effectiveness of the proposed algorithm.