A Self-adaptive Optimization Removal Algorithm for Continuum Evolutionary Structural Topology

This paper presents a new self-adaptive optimization removal algorithm for evolutionary structural optimization (ESO) method. The elements rejection based on this algorithm is determined by their own attributes (statistical distribution of element sensitivity) rather than comparing the sensitivities to manual selective parameters in the traditional removal process. Thus, it avoids the choices and adjustments of a series of parameters, involves fewer subjective factors, and makes the removal process more adaptive. Furthermore, a self-adaptive coefficient is introduced to make the element removal vary with the iteration process and the calculation more stable. Besides, this algorithm is modified by the local secondary statistical distribution of sensitivity to depress the influence made by sensitivity concentration and improve the effectiveness of element removal. The numerous studies of structure cases based on this modified removal algorithm show that: the final results are reliable, and the optimization period is much shorter, the adaptability and robustness for different model are also well. Keywords—removal algorithm, adaptive optimization, evolutionary structural topology