Different from using equivalent output injection term in sliding mode to reconstruct the unknown input, this paper introduces another approach to reconstruct the unknown input for a class of uncertain systems. Based on a kind of reduced-order observer, and by employing the estimate of output vector's derivative which is obtained through constructing high-gain observers as approximate differentiators, an algebraic approach is used to estimate the unknown input. The error between the real unknown input and the estimate of unknown input can be reduced to any degree by adjusting the gain parameter. Moreover, an extension to measurement noise case is investigated. Finally, the simulation results show that the proposed method is effective.
Keywords: high-gain observer, measurement noise, reduced-order observer, unknown inputs estimation