Automated approach for optimizing dynamic systems

Abstract

The optimal design of nonlinear dynamic systems can be formulated as a multicriteria optimization problem. On the basis of a multibody system model integral type objective functions are defined evaluating the dynamic behavior of the system under consideration. Multicriteria optimization methods reduce the problem to nonlinear programming problems which can be solved with standard algorithms like the SQP method. The gradients required for such an efficient optimization procedure are computed by solving, additional differential equations resulting from an adjoint variable approach. The whole design process can be highly automated by using computer algebra packages.