高速铣削TC4表面粗糙度预测模型研究∗

摘要

The factors influencing the workpiece surface roughness is complexity and uncertainty, the cutting parameters are one factor of ones that have great influences on machined surface quality which can be con-trolled artificially. In order to choose appropriate cutting parameters to get better machined surface quality, the paper design orthogonal experimental and establish TC4 Empirical regression prediction model and gener-alized regression neural networks ( GRNN) for prediction of surface roughness when high speed milling TC4 and compare the predicting error. Results show that the established GRNN prediction mode has better predic-tion precision which can be used to control the surface roughness dynamically.%零件表面粗糙度的影响因素具有复杂性和不确定性，切削参数是能够人为控制并对零件的表面质量有较大影响的因素之一。为了优选合适的切削参数以达到提高零件表面加工质量的目的，通过设计正交试验并在此基础上建立了钛合金TC4高速铣削表面粗糙度的GRNN广义回归神经网络预测模型和经验回归模型，对其预测误差进行了比较分析。结果表明：所建立的GRNN预测模型较回归预测模型有更高的预测精度，能够更好的对表面粗糙度进行动态控制。