基于Inception-BiLSTM的小样本刀具磨损状态识别研究  

Recognition of Tool Wear State Based on Inception-BiLSTM Under Few-Shot

作　　者：魏永合[1] 王耿 吴静远 WEI Yonghe;WANG Geng;WU Jingyuan(School of Mechanical Engineering,Shenyang Ligong University,Shenyang 110159,China)

机构地区：[1]沈阳理工大学机械工程学院,沈阳110159

出　　处：《组合机床与自动化加工技术》2024年第5期147-151,共5页Modular Machine Tool & Automatic Manufacturing Technique

基　　金：辽宁省应用基础研究项目(101300230)。

摘　　要：针对工业生产中故障数据不足难以准确进行故障诊断问题,以Inception模块为主体结构,结合双向长短时记忆网络(BiLSTM),提出了Inception-BiLSTM故障诊断方法,并用刀具磨损状态识别进行实验验证。首先,将振动信号通过连续小波变换(CWT)得到时频特征图,利用Inception网络对时频图进行特征提取;然后,使用全局平均池化(GAP)将特征向量降维;最后,使用BiLSTM提取数据信息,以识别刀具磨损状态。实验结果表明,在小样本条件下,该方法相较于对比方法对刀具磨损状态识别的准确率更高。Aiming at the difficulty of accurate fault diagnosis due to insufficient fault data in industrial production,the Inception-BiLSTM fault diagnosis method is proposed with the Inception module as the main structure and combined with the bidirectional long-short-term memory network(BiLSTM).The method is presented and validated experimentally with tool wear state recognition.First,the vibration signal is obtained by continuous wavelet transform(CWT)to obtain the time-frequency feature map,and the Inception network is used to extract the features of the time-frequency map;then,the feature vector is reduced in dimension using global average pooling(GAP).Finally,BiLSTM is used to extract data information to identify tool wear status.The experimental results show that under the condition of small samples,the accuracy of this method is higher than that of the comparison method in identifying the tool wear state.

关 键 词：INCEPTION 双向长短时记忆网络 刀具 状态识别 连续小波变换 小样本 

分 类 号：TH161[机械工程—机械制造及自动化] TG71[金属学及工艺—刀具与模具]