论文部分内容阅读
提出了一种基于BP神经网络和遗传算法(GA)的多工况离散变量结构优化设计方法,并对某斗轮堆取料机回转平台进行优化设计。该方法将多工况问题处理为多约束问题,利用正交试验法选择神经网络训练样本点,通过参数化有限元模型计算出各工况下的样本数据,建立起基于BP神经网络的回转平台数学模型,为遗传算法提供适应度函数,最后运用遗传算法完成寻优计算。结果表明,回转平台自重减轻13.8%,取得了满意的优化效果。
A structural optimization design method based on BP neural network and genetic algorithm (GA) for discrete variables with multiple working conditions was proposed. The rotary platform of a bucket wheel stacker and reclaimer was optimized. The method treats multi-conditions as multi-constraint problem, selects the neural network training sample points by orthogonal test method, calculates the sample data under each working condition through the parametric finite element model, and establishes the rotary platform based on BP neural network Mathematical model to provide fitness function for genetic algorithm, and finally use genetic algorithm to complete the optimization calculation. The results show that the weight of the slewing platform is reduced by 13.8%, and a satisfactory optimization effect is achieved.