钛合金超塑性变形过程中的高斯回归预测
Predication of Gaussian Regression of Titanium Alloy During Superplastic Deformation Process
胡 静, 杨 屹, 陈德平, 曾 斌
点击：5951次 下载：0次
作者单位：(四川大学 制造科学与工程学院, 四川 成都 610065
中文关键字：钛合金超塑性； 流变应力； 高斯回归技术； GP模型
英文关键字：superplasticity of titanium alloy; flow stress; Gaussian regress technology; GP model
中文摘要：在预设实验条件下，利用Gleeble-3500D热模拟机，完成了钛合金TC4高温超塑性拉伸试验。然后利用处理高度非线性问题的高斯回归技术，借助MATLAB语言编程，对高温超塑性拉伸过程中的流变应力进行了预测，其平均绝对误差0.67 MPa，平均相对误差2.91%。与神经网络预测结果相比，其预测精度更高且简单易行，是钛合金超塑性变形过程中参数预测和优化的可行工具。
英文摘要：Under the prepared experimental conditions, high temperature superplastic tensile experiments of TC4 were finished using Gleeble-3500D thermal simulation instrument. The flow stress of TC4 alloy was forecasted using of Gaussian regress technology which is appropriate for highly nonlinear problems, and with the help of the programme of MATLAB language. The results show that the mean absolute error is 0.67 MPa, the mean relative error is 2.91%. Compared with neural network, GP prediction accuracy is higher and simpler, it's good for forecasting and optimizing the parameters of superplastic deformation process of titanium alloy.