基于切削声和切削力参数融合的刀具磨损状态监测

针对单一传感器监测刀具磨损状态存在的不足,提出了将声传感方式和力传感方式综合利用,以人工神经网络作为多传感器信息融合的方法.在立式数控加工中心上铣削加工45<'#>钢调质试件,利用驻极体传声器和Kistler测力仪检测与刀具磨损相关的特征量,得出铣削声信号特征量LPCC的第6、7、8阶分量,X、Y向切削力以及绕z轴的力矩与刀具磨损密切相关.以这6个特征量作为神经网络的输入信号,利用有动量的梯度下降的BP算法建立了刀具磨损状态监测的多参数融合模型.研究结果表明神经网络输出值与实际测量值基本相符合,切削声和切削力特征融合后提高了识别刀具磨损程度的准确性和稳定性.

The Tool Wear Condition Monitoring Based on the Parameters Fusion of the Cutting Noise and the Cutting Force

For the shortcomings of single sensor monitoring tool wear condition,we put forward the method which sound and force sensing sensor are comprehensive utilized and artificial neural network is used as the multi-sensor information fusion.The characteristics associated with tool wear are studied in the vertical machining center using electret microphone and Kistler dynamometer.The experiment results show that the 6th,7th and 8th order components in the characteristic parameters LPCC of the milling sound signal...