NonLinear System Identification Using Cumulants

Abstract : In this paper we discuss the problem of nonlinearquadratic systems identification. The most existing methods, inthis area, are based on second order or higher order moments. Inmost cases, these methods are more complicated. In this paper wepresent a technique based on a minimum of equations to identifythe parameters of the system. The proposed algorithm is basedon cumulants. The simulation results show that the presentedalgorithm gives very satisfactory results, even in the case of asmall number of samples and in noisy environment.