The formulation of parameter identification in multibody dynamics

Abstract

Multibody dynamics is an effective method for dynamic analysis of mechanisms composed of multiple rigid and flexible bodies, and is used in many simulations. In recent years, it has become possible to simulate complex and multidegree-of-freedom mechanisms using multi-body dynamics. However. some mechanisms have parameters that can not be measured. In such cases, those parameters are estimated from measurable data. In previous researches proposeed the identification method for multibody system using adjoint method. The adjoint method is an approach for the computation of the gradient of a cost function to identify parameters. Incidentally, the equation of motion for multibody systems are described in differential algebraic equations (DAEs) consisting of differential equations and constraint equations. When solving the DAEs, it is common to use the second over time derivative of the constant equations. However, that method has a problem that the configuration level constraints are not properly maintained. It becomes a big problem for complex system. Therefore, this study proposed the parameter estimation method using adjoint method by solving the DAEs without using the second over time derivative of the constant equations. The equations of motion were discretized by the midpoint scheme and solved by the Newton-Raphson method. Two examples show that the correct parameters are found by the present parameter estimation method.