东北大学学报（自然科学版）

东北大学学报(自然科学版) ›› 2013, Vol. 34 ›› Issue (9): 1292-1295.DOI: -

• 机械工程 • 上一篇    下一篇

基于广义维数与优化BP神经网络的刀具磨损量预测

张锴锋1，袁惠群2，聂鹏3   

  1. (1东北大学机械工程与自动化学院，辽宁沈阳110819；2东北大学理学院，辽宁沈阳110819；3沈阳航空航天大学机电工程学院，辽宁沈阳110136)
  • 收稿日期:2012-12-11 修回日期:2012-12-11 出版日期:2013-09-15 发布日期:2013-04-22
  • 通讯作者: 张锴锋
  • 作者简介:张锴锋(1980-)，男，辽宁沈阳人，东北大学博士研究生;袁惠群(1954-)，男，河北石家庄人，东北大学教授，博士生导师.
  • 基金资助:
    国家高技术研究发展计划项目(2012AA040104).

Prediction of Tool Wear Based on Generalized Dimensions and Optimized BP Neural Network

ZHANG Kaifeng1， YUAN Huiqun2， NIE Peng3   

  1. 1. School of Mechanical Engineering & Automation， Northeastern University， Shenyang 110819， China; 2. School of Sciences， Northeastern University， Shenyang 110819， China; 3. School of Mechanical & Electrical Engineering， Shenyang Aerospace University， Shenyang 110136， China.
  • Received:2012-12-11 Revised:2012-12-11 Online:2013-09-15 Published:2013-04-22
  • Contact: YUAN Huiqun
  • About author:-
  • Supported by:
    -

RichHTML

0

PDF (PC)

592

摘要： 根据多重分形理论，采用改进的盒计数法计算了切削加工过程中声发射(AE)信号的广义分形维数，得到了不同刀具磨损状态下AE信号的广义维数谱，分析了广义维数与刀具磨损量之间的关系.以广义分形维数以及切削加工参数为特征，进行归一化处理后作为BP神经网络输入向量；采用遗传学算法，对BP神经网络的初始权值和阈值进行了优化，利用优化后的神经网络对刀具磨损量进行预测.测试结果表明，该方法可以较精确地预测刀具磨损量，平均预测误差为001mm.

关键词: 广义分形维数, BP神经网络, 刀具磨损, 预测, 遗传算法

Abstract: Based on multifractal theory，the generalized fractal dimensions of acoustic emission(AE)signals during cutting process were calculated using improved boxcounting method.The generalized dimension spectrums of AE signals under different tool wear conditions were obtained，and the relationship between tool wear quantity and generalized dimensions was analyzed.Together with cutting process parameters， the generalized fractal dimensions were taken as the input vectors of BP neural network after normalization.The initial weight and bias values of BP neural network were optimized with genetic algorithm which was used to predict the tool wear quantity.The test results show that the method can be effectively used for the prediction of tool wear， and the mean prediction error is 001mm.

Key words: generalized fractal dimensions, BP neural network, tool wear, prediction, genetic algorithm

中图分类号: