论文部分内容阅读
利用神经网络对有限元计算的样本数据建立起结构设计参数（输入）与位移、应力等（输出）的全局性映射关系，以获得遗传算法求解结构优化问题所需的目标函数值，从而进一步获得问题的优化解．5杆桁架算例表明，此方法可以在较少的有限元分析次数下获得较好的优化解．
Neural network is used to establish the global mapping relationship between structural design parameters (input) and displacement, stress (output) of the sample data of finite element calculation so as to obtain the objective function value needed by genetic algorithm to solve the structural optimization problem, so as to obtain further Optimal solution to the problem. The results of 5-truss truss show that this method can obtain a better optimal solution with fewer finite element analysis times.