基于主轴电流和振动信号的数控机床刀具磨损在线预测  

Online Prediction of Tool Wear of CNC Machine Tool Based on Spindle Current and Vibration Signal

作　　者：朱炜炜 杨赟 李全林 葛广言 王刚 杜正春[2] 杨建国 Zhu Weiwvei;Yang Yun;Li Quanlin

机构地区：[1]宁波科威联创数控技术有限公司,浙江宁波315400 [2]上海交通大学机械与动力工程学院,上海200240

出　　处：《机械制造》2023年第2期70-75,78,共7页Machinery

基　　金：宁波市制造业创新中心项目(编号:20213605)。

摘　　要：为了实现数控机床加工过程中刀具磨损状态的在线预测,提高数控机床智能化水平,提出一种基于主轴电流和振动信号的数控机床刀具磨损在线预测方法。这一在线预测方法采集能够反映刀具磨损状态的主轴电流和振动信号,对信号进行频域、时频分析处理,采用小波包分解和经验模态分解两种方法进行特征提取,得到与刀具磨损状态变化密切相关的特征值,按照递增或递减趋势进行保序回归操作,使用指数平滑方法进行平滑处理,由此建立基于遗传算法参数寻优的支持向量回归模型,用于预测刀具磨损量。试验及应用表明,应用这一在线预测方法,刀具磨损预测的平均误差在25μm以内,满足企业加工要求。In order to realize the online prediction of tool wear status in the machining process of CNC machine tool and improve the intelligent level of CNC machine tool,an online prediction method of tool wear of CNC machine tool based on spindle current and vibration signal was proposed.This online prediction method collects the spindle current and vibration signal that can reflect tool wear state,and analyzes the signal in the frequency domain and time frequency.The feature extraction is carried out by wavelet packet decomposition method and empirical mode decomposition method,and the feature value closely related to the change of tool wear state is obtained,and the sequential regression operation is carried out according to the increasing or decreasing trend,and the exponential smoothing method is used for smoothing,so as to establish a support vector regression model based on genetic algorithm parameter optimization to predict the tool wear amount.Tests and applications show that the average error of tool wear prediction is within 25μm by applying this method,which meets the machining requirements of enterprise.

关 键 词：主轴 电流 振动 数控机床 刀具 磨损 预测 

分 类 号：TG71[金属学及工艺—刀具与模具]