基于改进遗传算法和LVQ网络的刀具故障诊断  

Tool fault diagnosis based on improved genetic algorithm and LVQ network

作　　者：王文昊 李海伟[2] 聂鹏 王焕棋 张锴锋 Wang Wenhao;Li Haiwei;Nie Peng;Wang Huanqi;Zhang Kaifeng(College of Mechanical and Electrical Engineering,Shenyang Aerospace University,Liaoning Shenyang,110136,China;Shenyang Aircraft Industry(Group)Co.,Ltd.,Liaoning Shenyang,110136,China)

机构地区：[1]沈阳航空航天大学机电工程学院,辽宁沈阳110136 [2]沈阳飞机工业(集团)有限公司,辽宁沈阳110136

出　　处：《机械设计与制造工程》2022年第8期60-64,共5页Machine Design and Manufacturing Engineering

基　　金：辽宁省教育厅重点实验室资助项目(LS2010117);沈阳市人才资源开发专项基金资助项目(Syrc201007)。

摘　　要：针对叠层材料钻削加工特点,提出一种通过改进遗传算法(SAMGA)优化学习向量量化(LVQ)网络的刀具磨损在线监测方法。该方法在刀具磨损监测实验过程中,采集制孔过程中的声发射信号与红外温度信号,利用小波包分解与主元分析法对采集到的信号进行滤波与降维处理,将处理后的信号特征作为输入特征向量导入到LVQ网络模型中,并通过改进遗传算法优化其初始权值与阈值。结果表明:SAMGA-LVQ模型相比BP网络对于刀具磨损的预测识别精度更高,改进遗传算法对LVQ网络优化后训练速度有明显提升,更适用于刀具磨损在线监测系统。Aiming at the characteristics of lamination material drilling,this paper presents an improved genetic algorithm(SAMGA)optimized learning vector quantization(LVQ)network for on-line tool wear monitoring.The method in the course of tool wear monitoring experiment,hole of AE in the course of acoustic emission signal acquisition system and infrared temperature signal,the wavelet packet decomposition and principal component analysis is carried out on the collected signal filtering and dimension,and after processing the signal features as input vector of the imported into the LVQ network model,and using the improved genetic algorithm to optimize the first weights and sills.The consequence indicate that the SAMGA-LVQ model is more exact than BP network in the prediction and recognition of tool wear,and the improved genetic algorithm can significantly improve the training speed after optimization of LVQ network,so it is more suitable for the on-line monitoring system of tool wear.

关 键 词：钻削加工 刀具磨损 改进遗传算法 LVQ网络 故障诊断 在线监测 

分 类 号：TP206.3[自动化与计算机技术—检测技术与自动化装置]