采用不同工艺参数研究了激光熔凝K465高温合金的开裂行为, 并采用BP神经网络模型描述了裂纹指数与工艺参数的关系。研究发现, 激光熔凝K465合金的裂纹主要分布在熔池顶部和底部的界面处, 并呈现典型的液化开裂特征。通过建立激光熔凝区裂纹指数的数学描述方法, 以激光功率、扫描速度、光斑直径和预热温度作为输入参数, 以裂纹指数为输出参数, 发展了一个均方误差小于10-8的BP神经网络模型, 并可对实验结果进行较好的预测。

The crack behaviors were investigated in K465 superalloy by laser remelting using different processing parameters. The relationship between the crack index and processing parameter was described by BP neural network. The cracks were mainly distributed near the top and bottom boundary of the molten pool, and presented an intergranular liquation cracking pattern. The new crack index was defined to describe the character of the cracks in laser remelting K465 superalloy. Using the laser power, scanning velocity, beam spot diameter and preheating temperature as the inputs of model, and the crack index as the output of this model, the BP neural network is established with the mean square error less than 10-8. The predictions of BP neural network show a good agreement with the experimental results.