Application of Gradient Search Procedures for the Identification of Unknown System Parameters From System Response Observations

Abstract

Techniques for identifying and evaluating the physical parameters of nonlinear dynamic systems have been investigated A method has been developed for converting the measured dynamic response of a system into estimates of its unknown dissipation function. A computational algorithm employing an optimization or gradient search technique was written for determining the unknown system parameters. This procedure uses a least squares criterion function to determine the measure of agreement between the mathematical representation of the system and the physical system. A rather general statement of the nonlinear equation of motion allows a better final determination of the system than that provided by some fixed-form describing function. The technique is especially adaptable to the identification problem of determining the dissipation function of a nonlinear shock absorber.