Unknown input estimation by applying extended Kalman filter based on unknown but bounded uncertainties

In this paper a new unknown input estimation method is proposed for a class of nonlinear stochastic systems in the presence of time dependent unknown inputs, when the system states and process noises are unknown but bounded. In this study, a new augmented state vector is constructed by augmenting unknown inputs as a new state to the original state vector. Then a recursive algorithm based on unknown but bounded (UBB) uncertainty is developed, that unlike the Bayesian models which consider the state estimate as a single vector, produces a time-varying set of state estimates that contains the system's true state. The particular sets to be discussed, are ellipsoids. The proposed method doesn't need any unknown input detection stage procedure and covariance resetting that are necessary in previous works. At last, efficiency of the proposed method is shown in a numerical simulation for a nonlinear system.