A sequential refinement approach for parameter optimization in continuous dynamic models

The determination of the optimal values for parameters in a continuous dynamic system model is normally a computationally intensive task. Two separate numerical processes are involved; namely, the mechanism for solving the ordinary differential equations that comprise the system model, and the function minimization procedure used to search for the optimal parameter values. Both these processes typically have embedded parameters which control their respective operations. In this paper a general approach is described for adjusting these parameters in a way which allows the two processes to function in a more integrated and hence more efficient way in solving the parameter optimization problem. A specific implementation of the approach is described and the results of an extensive set of numerical experiments are given, These results indicate that the approach can provide a significant advantage in reducing the computational effort.