摘要
针对传统智能监测方法中信号特征提取繁琐、信息易丢失及易陷入局部最优等问题,本文提出一种基于深度学习理论的刀具磨损监测方法,用于刀具的量化磨损状况在线监测。该方法设计了卷积神经网络结构,给出了一种数据图片化处理方法,构建了刀具磨损量化监测模型。该方法有效避免了对信号特征提取与筛选的依赖,实现磨损特征信息的自适应提取和识别。实验结果表明,与传统神经网络相比较,本刀具磨损卷积神经网络监测模型监测过程简单,具有较高的监测精度,实现了大样本下刀具磨损特征的自适应提取与磨损值的在线量化监测,为实现制造系统中刀具的动态调度提供了基础。
In the traditional intelligent monitoring method,the problem of signal feature extraction is cumbersome,information is easy to lose and easy to fall into local optimum.This paper proposes a tool wear monitoring method based on deep learning theory,which is used for online monitoring of quantitative wear conditions of tools.In this method,the structure of convolutional neural network is designed,a data pictorial processing method is given,and a quantitative monitoring model of tool wear is constructed.The method proposed in this paper effectively avoids the dependence of signal feature extraction and screening,and realizes the adaptive extraction and recognition of tool wear features.The experimental results show that compared with the traditional neural network,convolutional neural network monitoring model for the tool wear has a simple monitoring process and high monitoring accuracy.The adaptive extraction of tool wear characteristics and online quantitative monitoring of wear values under large samples are realized,which provides a basis for realizing dynamic scheduling of tools in manufacturing systems.
作者
欧阳俊
李孝元
李少坤
Ou-yang Jun;Li Xiao-yuan;Li Shao-kun(Wuhan University of Engineering Science,Hubei Wuhan 430200)
出处
《内燃机与配件》
2025年第2期11-14,共4页
Internal Combustion Engine & Parts
关键词
刀具磨损监测
深度学习
卷积神经网络
特征提取
Tool wear monitoring
Deep learning
Convolutional neural network
Feature extraction
