基于BP神经网络模型的机床大件结构动态优化方法及其应用研究

将BP神经网络理论与有限元建模方法相结合，提出了采用BP神经网络建立机床整机主要部件的动力学模型，并应用大型有限元分析软件ANSYS的APDL进行BP神经网络样本的快速采样的方法，根据所提出的方法，建立了机床双W筋板床身的筋板位置，厚度与床身前5阶频率之间的BP神经网络模型，并以床身第1阶固有频率最高为目标进行了设计变量的自动搜索寻优计算且获得了满意的结果，表明神经网络理论与传统的数值方法相结合应用于实体结构的动态分析计算具有重要的现实意义。

Dynamic optimization of large parts of machine tool based on BP neural network model

Combined BP neural networks theory with the finite element modeling method, the dynamic models of the main parts on machine tool are presented. The fast sampling of BP neural network samples is carried out by using APDL of ANSYS. The BP neural network model corresponding to bar position, bar thickness and the preceding 5 frequency of body of double W bars is built based to the proposed method.To set the first frequency to be the largest, the automatic searching of the optimal solution of design variables is done, and the result is satisfactory, which shows it great reality meaning that the method combined neural network theory with numerical value method is applied to the analysis calculation of solid structures.