Abstract: According to the definition of surface topography,it could be separated into two parts of macroscopic shape error and microscopic surface roughness.An integrated method which included geometric modeling and neural network was proposed to simulate and predict surface topography.By using the principles of transformation figure matrix and vector operation,the movement path equation of ball-nose end mill versus work-piece was derived.A simulation model for the three-dimensional surface topography generated by a ball-nose end mill was established,which could predict the shape error.By means of MATLAB software,the Back Propagation (BP) neural network prediction model for surface roughness was established.The accuracy of the simulation and prediction models was verified by experiments,which indicated the effective prediction for surface topography.

Key words: surface topography, ball-nose end mill, shape error, surface roughness, simulation, prediction

摘要： 根据表面形貌的定义,将球头铣刀加工表面分离为宏观的形状误差和微观的表面粗糙度两部分,综合运用几何建模和神经网络对表面形貌进行仿真预测。利用图形矩阵变换原理和矢量运算法则,推导出球头铣刀相对于工件的运动轨迹方程,建立了基于MATLAB软件的三维表面形貌仿真模型对形状误差进行预测。借助于MATLAB软件,通过反复训练建立了BP神经网络表面粗糙度预测模型。通过实验验证了仿真预测模型的准确性,表明所建立的模型具有有效的预测作用。

关键词: 表面形貌, 球头铣刀, 形状误差, 表面粗糙度, 仿真, 预测