弹塑性力学问题的神经网络计算模型

Model for solving the elastoplasticity based on neural networks

为了提高弹塑性力学问题的计算速度 ,给出了其新的势能变分原理形式。利用神经网络的并行、分布处理的特点进行寻优 ,并给出了相应网络的结构形式和网络参数。

Parametric Variational Principle for Elastoplasticity was improved to adapt for the computation based on neural networks. The structure and parameters of the neural networks, which is used to solve the elastoplasticity are given. By using the method, the solution of the elstoplasticity can be obtained within an elapsed time of the circuit time\|constant(ns).