论文部分内容阅读
文章针对传统的机械零件多目标优化算法的不足,提出了一种基于改进型BP神经网络和NSGA-Ⅱ遗传算法的机械零件多目标优化设计方法,该方法首先利用Workbench对零件进行分析得到实验数据,然后用改进型BP神经网络对实验数据进行训练并建立起多目标优化的模型,采用NSGA-Ⅱ遗传算法对模型进行多目标优化。结果表明,在满足优化零件使用条件的情况下,运用该方法求得质量的相对误差最大为11%,变形的相对误差最大为3.36%,验证了该方法的有效性和可靠性。并将该方法得出的结果与传统Workbench得出的多目标优化结果进行了比较,证明了该方法优于传统Workbench优化方法。
In this paper, aiming at the deficiency of the traditional multi-objective optimization algorithm of mechanical parts, a multi-objective optimization design method of mechanical parts based on improved BP neural network and NSGA-Ⅱ genetic algorithm is proposed. The method firstly analyzes the parts by using Workbench to get the experimental data Then, the improved BP neural network is used to train the experimental data and a multi-objective optimization model is built. The model is optimized by NSGA-Ⅱ genetic algorithm. The results show that the relative error of quality is 11% and the maximum relative error of deformation is 3.36%. The validity and reliability of the proposed method are verified. The results obtained by this method are compared with those obtained by traditional Workbench. The results show that this method is superior to the traditional Workbench optimization method.