基于软计算的结构优化设计

【作者】 王李宁；

【导师】 郭海丁；

【作者基本信息】 南京航空航天大学 ， 机械设计及理论， 2003， 硕士

【摘要】 本文应用软计算方法中神经网络、遗传算法进行结构优化设计。针对神经网络在机构优化设计中网络结构难以确定的问题，进行了神经网络在结构优化设计中的映射性能和训练特性的研究；以此为基础，构造了一种适于结构优化设计的结构自适应BP神经网络，使得BP神经网络具有了一定的自我完善能力。利用该网络的快速映射能力代替结构优化设计中的有限元计算，并利用遗传算法作为搜索策略进行全局搜索寻优，成功实现了某型航空发动机涡轮盘的结构优化计算，取得较好的效果。论文还对连续体结构优化问题的样本选取问题进行了初步的分析，提出了一种样本优化的方法。更多还原

【Abstract】 Some methods of soft computing were used in the field of engineering structural optimization, including BP (Back Propagating) neural network and genetic arithmetic . Aiming at the problem that it is difficult to find a suitable structure of neural network in the structural optimization, the author conducted some researches on the mapping ability and training properties of BP neural network in the structural optimization and constituted a self-adapting BP neural network. The network was used to take the place of FEM (Finite Element Method) for it’s fast mapping ability in structure optimization and the genetic arithmetic was used as the searching method to accomplish global optimization. A turbine disk of areo-engine was optimized successfully with this method. Beside, the author has analyzed the way of getting suitable training samples, and carried out a method in which the training samples were optimized.更多还原