32th Congress of the International Council of the Aeronautical Sciences

04.2 - Aerostructures Design, Structural Dynamics, Aeroelasticity

A DYNAMIC SELF-CONTROL ERR STRATEGY FOR BI-DIRECTIONAL EVOLUTIONARY STRUCTURAL OPTIMIZATION

H. Zhang╣, Q. Zhang╣; ╣Sun Yat-sen University, China

The element removal ratio is a crucial parameter in bi-directional evolutionary structural optimization , which is of significance for the stability and efficiency of the optimization process and the final topology optimization results. The traditional BESO method uses a fixed evolution rate (the ERR method), which may ruduce inefficiency. And the existing self-updating evolution strategy ignores the influence of the initial element removal ratio and regulates the evolution rate to a higher level, which results in more potential unstability. In view of the above, this paper uses the relative standard deviation (RSD) representing the disperesion degree in statistics to define structural performance index , measures the uniformity of the structural element sensitivity, and analyzes the influence of the initial element removal ratio and the ideal change law of ERR in the process of structural optimization. Therefore, BESO is divided into two processes, the fast evolution phase and the smooth evolution phase, and it switches automatically from the fast one to the other according to the PI. The relative index of the structural performance fusing all iterative information is used as the control factor.By this control factor, this paper constructs the self-control function of ERR under two stages separately, which control the change of the evolution rate in real time.This method can obtain a faster optimization process on a stable basis. Finally, the implementation process of the BESO method with dynamic self-control ERR is put forward. The comparison of the results of related calculation examples shows that the method in this paper improves the optimization efficiency while taking into account its stability under similar optimization quality.


View Paper