IDENTIFICATION OF A CLASS OF NONLINEAR CONTINUOUS-TIME SYSTEMS USING HARTLEY MODULATING FUNCTIONS

Most of the existing approaches to identification of nonlinear dynamic systems involve matching a given input-output behaviour with empirica l discrete-time approximations such as artificial neural networks, Kol mogorov-Gabor polynomials, radial basis function networks, etc. Techni ques for dealing with physically-based continuous-time models are eith er applicable to only a restricted class of systems or are computation ally very demanding. In this paper a new methodology is presented that is applicable to a large class of nonlinear continuous-time systems, by defining a set of Hartley modulating functions for characterizing t he continuous process signals. The advantages of this new class of mod ulating functions are that a set of algebraic equations with real coef ficients results, the formulations are free from boundary conditions, and the computations can be made using fast algorithms for the discret e Hartley transformation. The resulting estimation scheme is applied t o different categories of nonlinear systems and is tested under both n oise-free and noisy conditions.