选区激光熔化316L不锈钢残余应力优化研究

利用人工神经网络的高度非线性函数逼近能力和自适应学习能力,建立选区激光熔化316L不锈钢成形残余应力预测模型。采用响应面法进行试验设计得到不同的工艺参数组合作为训练样本,通过残差分析验证样本数据的有效性,结合遗传算法对BP神经网络模型进行优化。结果表明,GA-BP神经网络模型的预测值与实际值的整体预测平均相对误差在5%左右,证明通过神经网络模型实现选区激光熔化316L不锈钢成形残余应力预测是可行的。

Using the highly nonlinear function approximation ability and adaptive learning ability of artificial neural network,a prediction model of residual stress in selective laser melting 316L stainless steel was established.The response surface method was used to design the experiment,and different process parameter combinations were obtained as training samples.The validity of the sample data was verified by residual analysis,and the BP neural network model was optimized with genetic algorithm.The results show that the overall average relative error between the predicted value of GA-BP neural network model and the actual value is about 5%,which proves that it is feasible to predict the residual stress of 316L stainless steel formed by selective laser melting through the neural network model.