Identification of frequency domain models for nonlinear systems

Abstract

This paper presents a computer-aided engineering approach for identification of linear models from a set of frequency response data. The approach is based on a new system identification technique. The primary intention of the developed system identification technique and the associated software is to identify linear models for nonlinear systems whose input/output behaviour is characterized by their corresponding sinusoidal-input describing function models. However, the technique may also be applied to identification of linear models from experimental frequency response data. At present, the identification approach and the associated software is limited to single-output, linear, deterministic, and time-invariant systems. A computer-aided engineering environment based on the developed system identification technique has also been developed. The software is developed on a Harris-800 super-minicomputer and a Tektronix 4115B high resolution, raster, and color graphics terminal.