Linear System State Estimation: A Neurocomputing Approach

A neurocomputing approach is developed in this paper to solve the problem of state estimation for linear dynamical systems. Dynamic optimization techniques are used to develop the adaptation laws for assigning the weights of a Hopfield net. Simulation results show that the new approach performs similar to Kalman filter, and outperforms it for some special situations. The new approach is very attractive for the real-time implementation of a state estimator for large scale systems.