TOWARDS REAL-TIME CONTINUOUS SYSTEM IDENTIFICATION USING MODIFIED HOPFIELD NEURAL NETWORKS

摘要

Solution of system identification problems on parallel analog networks is proposed. Recurrent neural networks whose dynamic equations have a Lyapunov function, relax to an equilibrium which is the minimum of the Lyapunov function. System identification problems are formulated in terms of a Lyapunov function and thus are solved using the recurrent networks. Convergence for linear and a set of nonlinear problems is assured. Identification of a simulated nonlinear system and of an experimental lightly damped structure demonstrate the practicality of the approach. Results show extremely fast solution times which are independent of the size of the problem.