This paper proposes a novel second-order predictive disturbance observer for systems experiencing external disturbance with both input and output measurement delays. The proposed observer uses the delayed system state and its derivative information to predict and compensate for future disturbance in real time, despite the delayed control input. Compared with the past studies about second-order disturbance observer, this paper analyses the impact of the disturbance derivatives on the estimation accuracy. This observer is designed using the Lyapunov–Razumikhin theorem and its exponential stability is proven. Comparative analysis and simulations demonstrate its advantages over second-order disturbance observers that do not consider delays.
Read full abstract