Optimal state and fault estimation for two-dimensional discrete systems

An optimal state and fault estimation scheme is proposed for two-dimensional discrete systems subject to either deterministic disturbances or stochastic disturbances (noises). A direct solution to the deterministic estimation problem is obtained first, based on a well-designed regularized least squares problem with a dynamic constraint of a two-dimensional singular system, augmented from the original system state and unknown disturbance. After proving the solution equivalence between the deterministic scenario and the stochastic one in the sense of optimal state and fault estimation, a unified solution, based on a Riccati-like equation recursion, can be established by weighting parameterization for two-dimensional systems in deterministic and stochastic cases. The unified solution also works as the optimal state observer and generalized Kalman filter for two-dimensional singular systems. Generalization discussions concerning different system descriptions with respect to fault as well as the implementations of the proposed estimator are also presented. Simulation illustrates the effectiveness of the proposed method.