基于尺度共轭梯度神经网络的TC4钛合金磨削烧伤预测  

Burns prediction of TC4-Ti-alloy based on scaled conjugate gradient neutral networks

作　　者：邓朝晖[1,2] 肖蓝湘[1,2] 邓辉[1,2] 刘伟[1,2] 

机构地区：[1]湖南科技大学难加工材料高效精密加工湖南省重点实验室,湖南湘潭411201 [2]湖南科技大学智能制造研究院,湖南湘潭411201

出　　处：《金刚石与磨料磨具工程》2017年第2期31-35,40,共6页Diamond & Abrasives Engineering

基　　金：国家科技支撑计划课题"机床主轴和船舶凸轮轴智能制造的工艺软件和知识库研发(2015BAF23B01)";湖南省研究生科研创新基金项目(CX2015B483)

摘　　要：为预测TC4钛合金高速外圆磨削烧伤的程度,基于磨削烧伤后组织相变导致的工件表面硬度值变化,以表面硬度值来反映工件的磨削烧伤程度。采用尺度共轭梯度算法的神经网络对TC4钛合金高速外圆磨削的表面硬度值进行预测,根据表面硬度值与磨削表面烧伤程度的对应关系预测磨削烧伤。验证实验表明:预测结果与实验结果的误差在5%以内,模型预测效果良好。In order to predict the degree of TC4 titanium alloy after high-speed cylindrical grinding, surface hardness is used to differentiate grinding burns of the workpiece based on surface hardness value change resulted by phase transformation after grinding burn. Surface hardness of TC4 titanium alloy after high-speed cylindrical grinding is forecasted using the scaled conjugate gradient algorithm of neural network. Correspondence relationship between the surface hardness value and the degree of grinding burn is used to predict grinding burns. Validation experiment indicates that the error between the experiments and the predictions is within 5%, which means that the model prediction effect is good.

关 键 词：TC4钛合金 表面硬度 烧伤预测 共轭梯度法 

分 类 号：TG58[金属学及工艺—金属切削加工及机床]