基于BP神经网络的矩形花键滚刀齿形设计

Simulation and Design of Rectangle Spline Hob's Profile Based on BP Neural Network

针对矩形花键滚刀的齿形设计问题,提出了应用BP神经网络逼近滚刀切削刃形的设计方法;基于BP神经网络的非线性映射特性,建立了滚刀理论齿形坐标点与模拟齿形曲线之间的BP网络模型。模拟仿真结果表明,此方法根据设置的目标误差逼近滚刀理论齿形,达到了精确设计矩形花键滚刀齿形的目的;与三点共圆法比较,神经网络方法大幅度提高了滚刀刃形的设计精度和设计效率。

Aiming at the design of rectangle spline hob＇s profile, the simulation method of using BP neural network to design cutting edge of rectangle spline hob is presented. Based on BP neural network＇ s nonlinear mapped characteristic a model of neural network between the theoretical profile and the simulation curve of a hob is established. The simulation experiment indicates that the method can exactly approach the hob＇s theoretical profile according to setting goal error. Compared with the method of the three - point build a circle, the simulation method of BP neural network can strongly increase both design precision and design efficiency.