基于卷积神经网络的铣刀后刀面磨损状态预测方法研究  

Research on Prediction Method of Milling Tool Flank Wear Condition Based on Convolutional Neural Network

作　　者：杨博宇 曹忠[1] 郭国强[1] 刘世民 沈彬[3] Yang Boyu;Cao Zhong;Guo Guoqiang;Liu Shimin;Shen Bin(Shanghai Spaceflight Precision Machinery Institute,Shanghai 201600,China)

机构地区：[1]上海航天精密机械研究所,上海市201600 [2]东华大学 [3]上海交通大学

出　　处：《工具技术》2022年第5期22-26,共5页Tool Engineering

基　　金：国家重点研发计划(2020YFB2010603)。

摘　　要：为了监测高温合金材料加工时的铣刀后刀面磨损状态,提出了基于卷积神经网络的刀具磨损状态预测方法,建立了基于机床主轴电流与功率信号实时监测的刀具磨损状态预测系统。通过建立与机床数控系统的通信,采集加工过程中的电流和功率信号,采用主成分分析法(PAC)对采集的参数进行特征提取,选择对刀具磨损值影响较大的主成分作为卷积神经网络的输入,实现对刀具磨损状态的准确预测。铣削实验结果表明,该方法具有较高的预测准确率。In order to monitor the flank wear condition of milling tool during the processing of superalloy materials,a tool wear condition prediction method based on convolutional neural network is proposed,and a tool wear condition prediction system based on real-time monitoring of spindle current and power signals is established.The current and power signals in the machining process are collected by communicating with the CNC system of the machine tool,and the characteristics of the parameters are extracted by the principal component analysis method(PAC).The principal component which has a great influence on the tool wear value is selected as the input of the convolutional neural network,so as to realize the accurate prediction of the tool wear condition.The results of milling experiments show that this method has a high prediction accuracy.

关 键 词：高温合金 刀具磨损 刀具状态预测 主成分分析 卷积神经网络 

分 类 号：TG61[金属学及工艺—金属切削加工及机床] TG132.3[一般工业技术—材料科学与工程]