A New Optimization Algorithm for Inverse Analysis to Material Parameters

Abstract

the inverse analysis to material parameters is often translated into an optimization for an objective function, based on the correlation between the material parameters and the foregone information. But mostly because of the non-linear correlation, a good optimization algorithm with the capabilities to avoid being trapped by local optima is required during the process of optimization. So the present paper proposes a new global optimization algorithm, which couples the dynamic canonical descent algorithm and the improved Powell’s algorithm. The high efficiency of the new algorithm is shown on four known problems classically for testing optimization algorithms and finally, in the non-linear inverse analysis, the new algorithm is used for optimizing an objective function to get material parameters rightly.