Examination of optimal design method using die press model (problem 25)
Various optimization methods, such as the direct search method, the simulated annealing method (SAM) and the genetic algorithm (GA) recently applied to the optimal design of magnetic devices, are investigated. The well-known global optimizers SAM and GA are used to find the global minimum of objective function. The effect of various parameters of optimization methods, for example, temperature parameter of SAM on the process of reaching the global minimum or local minimum was not investigated using actual magnetic devices. Therefore, the criterion for selecting optimal values of these parameters was not clear. In this paper, factors affecting convergence of various optimization methods, such as initial condition, temperature parameter of SAM, etc., are presented using the die press model. The local minimum and global minimum of the objective function for the die press model are shown in detail, and the properties of convergence process of the Rosenbrock's method (RBM), SAM and GA are examined. It is shown that SAM is better than GA and RBM in the case of the model discussed here. Two kinds of combination methods \((\text{SAM} + \text{RBM})\) and \((\text{RBM} + \text{SAM})\) are compared. It is shown that \((\text{SAM} + \text{RBM})\) is not always better than \((\text{RBM} + \text{SAM})\).
This page was built for publication: Examination of optimal design method using die press model (problem 25)
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4255037)
