神经网络在结构优化中的应用

Structure optimization with the neural network

采用阶梯折算及传递矩阵的基本方法,利用神经网络算法中的模拟退火法,研究了轴向力作用下非均匀变截面柱的屈曲优化问题,计算了给定柱体总体积约束下,使轴向受压柱的屈曲载荷达到最大的问题。研究表明:优化结果使4种条边界条件下的屈曲载荷都得到提高。

Using ladder reduction and transmission matrix method, buckling optimal problem of nonhomogeneous variable cross-section column compressed by axial force on four kinds of boundary conditions with isometric comfinement is solved in the paper. Buckling load increases, and simulated annealing method which is belong to neural network is used in this paper.