TC4合金热等静压热处理力学性能的BP人工神经网络预测

Prediction of mechanical properties of TC4 Alloy by hot isostatic pressing using BP artificial neural network

对激光选区熔化成形的TC4合金,进行不同工艺下的热等静压热处理试验,并对热处理后的试样进行室温抗拉强度、延伸率测试。以上述试验数据为基础,采用人工神经网络技术构建了TC4双相钛合金热等静压热处理温度、保温时间、压力为输入变量,室温抗拉强度、延伸率为输出变量的三层BP人工神经网络模型。通过对该模型的隐含层数、神经元个数、输入输出数据、算法函数进行选择与优化,设定预测精度,归一化输入输出参数,实现了对TC4合金不同热等静压热处理工艺参数下的力学性能的预测。

The hot isostatic pressing(HIP)tests were carried out on TC4 alloy formed by selective laser melting under different processes,and the tensile strength and elongation of the samples after heat treatment were tested at room temperature.Based on the above experimental data,a three-layer BP artificial neural network model was established by using artificial neural network technology.The input variables were temperature,holding time and pressure,and the output variables were tensile strength and elongation at room temperature.By selecting and optimizing the number of hidden layers,number of neurons,input and output data and algorithm function of the model,the prediction accuracy was set and the input and output parameters were normalized.The mechanical properties of TC4 Alloy under different hip heat treatment parameters were predicted.