基于神经网络的界面裂纹尖端场分析

通过构造反向传播神经网络，对裂纹尖端的应力场进行模拟，进而实现对裂纹尖端应力场甬数的逼近。得到的网络具有较高的联想、记忆能力和相当的稳定性，并且可以快速、准确地得到带裂纹构件的裂纹尖端应力场，从而确定裂纹尖端的塑性区和分析裂纹的扩展。数值计算给出了LY12-CZ材料裂纹扩展方向的计算结果，与实验结果吻合较好，还给出了两相材料含界面裂纹在复合型载荷作用下的塑性区形状的变化情况，并对两相材料含界面裂纹在复合型载荷作用下裂纹的扩展方向进行了预测。

Field Analysis of Interface Cracktip Based on Neural Network

The constructed Back Propagation Neural Network can simulate the stress field around the tip of the crack. Then the approximation of stress field around the tip of crack was achieved. The trained neural network has the ability of association and memory. At the same time, it can calculate the stress field of the component with a crack in reality quickly and correctly. By the way of Neural Network, plastic regions of crack tip can be obtained and the crack initiation was analyzed. The directions of crack extension are given for material LY12-CZ by numerical method. This results are similar to measured in the experiments. The directions of crack initiation have been predicted for interface crack subjected to mixed mode loading in bimaterials.