模具钢铣削中刀具磨损的试验研究

以面铣刀刀片磨损为研究对象,结合类神经网络系统建构高速数控铣削加工的预测模型。以加工参数为模型输入条件,刀腹磨耗为输出条件。采用多因素试验方法,选择切削速度、进给速度、切削深度三个试验参数,利用直交表式的试验计划法设计试验点。依照试验点铣削工件后再测量刀具加工后的刀腹磨耗量,进而求得倒传递网络所需的36组训练范例与11组验证数据。刀腹磨耗预测模式是利用类神经网络中的倒传递网络原理,以田口法求得倒传递网络参数的最优值。试验结果显示,刀腹磨耗随着切削速度、进给速度、切削深度增加而上升。铣削模具钢后,刀具磨耗预测值的平均误差为4.72%,最大误差为11.43%,最小误差为0.31%。整体而言,类神经网络对于铣削加工可进行有效预测。

Experimental Study of Tool Wear in Milling of Mold Steel

In face milling cutter blade wear for the study, combined with neural network system to construct high - speed CNC milling machining prediction model to the processing parameters for the model input conditions, flank wear as output condition. Multi - factor test method, select the cutting speed, feed rate, cutting depth of three experimental param- eters, the use of experimental method orthogonal array design experiments point type, the measurement tool flank wear vol- ume processed in accordance with the experimental points after milling a workpiece, thus obtained 36 sets of training exam- ples backpropagation network with 11 groups needed to validate data. Flank wear prediction model using neural network back - propagation network theory, in order to obtain the best value Taguchi method backpropagation network parameters. Experimental results show that flank wear with cutting speed, feed rate, cutting depth increases to rise. After milling tool steel, tool wear prediction average error value is 4.72% , the maximum error is 11.43% , the minimum error of 0.31%. Overall, the neural network can effectively predict the milling process.