A novel neurocomputing approach to nonlinear stochastic state estimation

This paper presents a novel neuro-computing approach to the problem of state estimation by means of an hybrid combination of Hopfield neural network whose capability of solving certain optimization problems is well-known and feedforward multilayer neural net which is very popular because of its universal approximation property. This neuro-estimator is very appropriate for the real-time implementation of linear or/and especially nonlinear state estimators. Simulation results shows the effectiveness of the proposed method.