类椭圆吊臂结构优化设计与研究

Optimization and research of a similar-oval crane arm based on BP neural network and genetic algorithm

吊臂轻量化是进行吊臂结构优化设计的主要目标,但吊臂优化设计存在非线性程度高的特点,传统的优化方法很难得到全局最优解。首先建立了类椭圆吊臂有限元分析的参数化仿真模型,采用正交试验法建立样本数据,完成对BP神经网络的训练,建立了设计参数与目标向量之间的非线性映射关系。最后以吊臂重量最轻为优化目标,利用遗传算法寻优。优化结果表明,吊臂自重显著降低。该方法为以后复杂结构的轻量化设计提供了一种新的思路。

Lightweight is the main target of the optimization of the crane arm. Nevertheless, there existing high nonlinearity during process of the optimization. It＇ s difficult to acquire the best result under the whole situation by means of the traditional optimiza tion methods. Initially,the parameterized simulative model of the similar-oval crane arm is established,then an orthogonal experi- mentation was used in choosing the training sample data and trained by the BP neural network, the nonlinear neural network mod- el between the design parameters and the objectiyes is constructed based on the test data. Aiming at the least weight of the crane arm, the genetic algorithm was used as to accomplish the optimization finally. The optimization results indicate that the crane arm＇ s weight descends obviously. This method supply a new way for the later lightweight designs of complicated structure.