Adaptive distributed unknown input observer for linear systems

Abstract

This paper studies the adaptive distributed unknown input observer (ADUIO) for linear systems with local outputs, which contains a group of local observers under directed graph. The difficulty is the adaptive estimation of global output for the systems with unknown inputs. To solve the problem, disturbance decoupling principle and leader-following consensus strategy are integrated to estimate local outputs of other observers, which can be accumulated to recover the global outputs. Based on the estimated global outputs, an ADUIO is constructed to estimate the full state which avoids using local output matrices of other observers and global information of the graph. Different from the extensive joint detectability assumption in existing results, a detectability assumption is given to make the estimation errors converge to zero asymptotically.