Parameter Identification in Multibody Systems Using Lie Series Solutions and Symbolic Computation

Abstract

This paper studies the application of the Lie series to the problem of parameter identification in multibody systems. Symbolic computing is used to generate the equations of motion and the associated Lie series solutions automatically. The symbolic Lie series solutions are used to define a procedure for computing the sum of the squared Euclidean distances between the true generalized coordinates and those obtained from a simulation using approximate system parameters. This procedure is then used as an objective function in a numerical optimization routine to estimate the unknown parameters in a multibody system. The effectiveness of this technique is demonstrated by estimating the parameters of a structural system, a spatial slider-crank mechanism, and an eight-degree- of-freedom vehicle model.