In this thesis, a method of systematic identification of unknown dynamic parameters of constrained mechanical systems, through application of optimization technique, is presented.
The identification problem is treated as an optimization problem, by defining the error function between the real and modelled systems as cost function while nonlinear algebraic kinematic constraint equations and differential equations of motion of the systems as state equations. Unknown dynamic parameters are chosen as design variables. Parameter identification is achieved by reducing the cost function in the feasible domain of the problem.
An adjoint variable technique for the first order design sensitivity analysis of functions in the problem and the gradient projection method for optimization are utilized.
A computer program based on the theory is developed and tested with several example problems. Research results demonstrate that the method presented has feasibility of further development and application.