|
| Bi-directional evolutionary structural optimization method based on control for evolutionary step length of element density |
| Received:September 24, 2014 Revised:November 19, 2014 |
| View Full Text View/Add Comment Download reader |
| DOI：10.7511/jslx201601003 |
| KeyWord:evolutionary structural optimization method evolutionary step length of element density weight coefficient sensitivity error checkerboard pattern |
| Author | Institution |
| 匡兵 |
桂林电子科技大学 机电工程学院, 桂林 |
| 李应弟 |
桂林电子科技大学 机电工程学院, 桂林 |
| 刘夫云 |
桂林电子科技大学 机电工程学院, 桂林 |
| 刘娟 |
桂林电子科技大学 机电工程学院, 桂林 |
|
| Hits: 1968 |
| Download times: 1433 |
| Abstract: |
| In evolutionary structural optimization (ESO) method, evolutionary step length of element density is one of key factors to obtain global optimal solution of optimization problem.To improve the ability of global optimization for ESO method, this paper presents bi-directional evolutionary structural optimi-zation method (BESO) based on control for evolutionary step length of element density.Control model of evolutionary step length of element density is built according to weight coefficient of each unit impacting on structure performance.The rate of elements being deleted or added of the principal elements and the secondary elements is controlled by using the proposed control model.Evolutionary step length of element density is controlled so that error of sensitivity can be decreased and gray elements are controlled.To avoid optimization of failure on account of error deleting element, the operation for add element of BESO method is combined based on controlling evolutionary step length of element density.At the same time, checkerboard pattern is controlled by using sensitivity redistribution technology to obtain smooth optimal structure.Finally, the presented method is verified by two numerical examples can effectively decrease sensitivity error by controlling evolutionary step length of element density to improve the ability of global optimization. |