赵书东,禹晓敏,王文玺,邹莱.金字塔砂带磨损状态的声信号GA-BP识别方法[J].表面技术,2024,53(3):28-38.
ZHAO Shudong,YU Xiaomin,WANG Wenxi,ZOU Lai.GA-BP Identification of Acoustic Signals for Wear States of Pyramidal Abrasive Belts[J].Surface Technology,2024,53(3):28-38
金字塔砂带磨损状态的声信号GA-BP识别方法
GA-BP Identification of Acoustic Signals for Wear States of Pyramidal Abrasive Belts
投稿时间:2023-10-27  修订日期:2024-01-18
DOI:10.16490/j.cnki.issn.1001-3660.2024.03.003
中文关键词:  机器人砂带磨削  声信号  Archard模型  遗传算法优化BP神经网络
英文关键词:robotic abrasive belt grinding  acoustic signal  Archard model  genetic algorithm optimized BP neural network
基金项目:国家自然科学基金青年科学基金项目(52105430);中国博士后科学基金面上项目(2023M740398);重庆市自然科学基金创新群体项目(cstc2019jcyj-cxttX0003);中央高校基本科研业务费资助(2023CDJXY-024)
作者单位
赵书东 重庆大学 机械与运载工程学院 重庆 400044 
禹晓敏 中国航发航空科技股份有限公司,成都 610599 
王文玺 重庆大学 机械与运载工程学院 重庆 400044;重庆大学 高端装备机械传动全国重点实验室,重庆 400044 
邹莱 重庆大学 机械与运载工程学院 重庆 400044;重庆大学 高端装备机械传动全国重点实验室,重庆 400044 
AuthorInstitution
ZHAO Shudong College of Mechanical and Vehicle Engineering,Chongqing 400044, China 
YU Xiaomin AECC Aero Science and Technology Co., Ltd., Chengdu 610599, China 
WANG Wenxi College of Mechanical and Vehicle Engineering,Chongqing 400044, China ;State Key Laboratory of Mechanical Transmissions for Advanced Equipment, Chongqing University, Chongqing 400044, China 
ZOU Lai College of Mechanical and Vehicle Engineering,Chongqing 400044, China ;State Key Laboratory of Mechanical Transmissions for Advanced Equipment, Chongqing University, Chongqing 400044, China 
摘要点击次数:
全文下载次数:
中文摘要:
      目的 金字塔砂带连续磨损会引发钝峰、材料去除能力差和产热多等问题,为避免砂带磨损造成加工效率持续降低和工件表面质量逐渐恶化,需提高金字塔砂带磨损预测能力。方法 在配有声音采集系统的力控机器人磨削系统中对钛合金工件进行了砂带磨损试验;基于Archard模型建立了金字塔砂带磨损模型,并对金字塔砂带磨损程度进行量化;然后利用短时傅里叶和小波包分解分析、提取砂带磨损相关的声音特征;基于声音信号特征建立GA-BP模型,并对金字塔砂带磨损状态进行预测。结果 Kr与R0规律相近,随着磨削速度的增大而略微增大。对磨削声音进行小波包分解,DD2频段的声音特征随磨削时间逐渐降低,相较于其他频段更具有规律性。提取DD2频段的声音信号特征建立GA-BP模型,并对金字塔砂带磨损状态进行预测。结果表明,决定系数(R2)大于0.8,平均绝对误差(MAE)小于0.04,平均偏差误差(MBE)在±0.002之间,均方误差(RMSE)小于0.05。结论 随着砂带的磨损,金字塔尖锐的胞体开始磨平,单颗胞体的局部压力逐渐减小,材料去除能力减弱,产生的微振荡越来越弱,高频信号的声音特征逐渐下降。通过DD2频段声音信号特征建立的GA-BP模型对金字塔砂带磨损状态进行预测,具有准确性和稳定性。
英文摘要:
      Continuous wear of pyramid belts causes problems such as blunt peaks, poor material removal ability and high heat generation, which is particularly obvious in the grinding and polishing of materials such as high temperature alloys and titanium alloys. In order to avoid the phenomena of continuous reduction of processing efficiency and gradual deterioration of workpiece surface quality caused by belt wear, the prediction capability of pyramid belt wear needs to be improved. Experiments were conducted by a robotic belt grinding system and a brand new 237AA pyramid belt manufactured by 3M. The full-life pyramid belt wear experiments were conducted on titanium alloy workpieces at three different grinding speed in a dry grinding condition. The grinding sound of the abrasive belts was captured by a microphone at a transverse position located 4 cm from the grinding surface. Based on the mathematical derivation of Archard model, it was proposed to quantify the wear degree of pyramid abrasive belts in terms of Rat, and a pyramid abrasive belt wear model was obtained. Then, the frequency distribution and amplitude change of idling sound and grinding sound of abrasive belt in different wear periods were obtained by short-time Fourier to analyze the sound signals. The frequency bands with correlation with the degree of wear of abrasive belt were obtained by decomposing the wavelet packet of the original signals and extracting the features. Finally, a GA-BP model was established based on the sound signal features to predict the wear state of the pyramid abrasive belt. Kr and R0 were obtained by fitting the wear Rat. Kr was related to the characteristic parameters of the pyramid belt and characterized the wear rate of the belt cone. The difference of Kr under different speed was small, but it increased slightly with the increase of speed. R0, as the initial Rat of pyramid abrasive belts, also showed a similar law with Kr. By performing short-time Fourier analysis and wavelet packet decomposition on the grinding sound, it could be obtained that the frequency of the idling sound was mainly concentrated in the low frequency band. The grinding sound in different wear periods of abrasive belts was distributed in all frequency bands, and the sound in the low-frequency band had a similarity with the frequency distribution of the idling sound of abrasive belts. The sound characteristics of the DD2 frequency band gradually decreased with the grinding time, which was more regular than the other frequency bands. The results showed that the coefficient of determination (R2) was greater than 0.8, the mean absolute error (MAE) was less than 0.04, the mean deviation error (MBE) was in the range of ±0.002, and the mean square error (RMSE) was less than 0.05. Rat correlates extremely well with the material removal capacity of pyramid belts and accurately quantifies the degree of belt wear. As the abrasive belt wears, the sharp pyramidal cones begin to flatten out, the localized pressure of a single cone gradually decreases, the material removal capability weakens, the micro-oscillations generated by the abrasive belt to remove material become weaker and weaker, and the acoustic signature of the high-frequency signal gradually decreases. Acquisition of sound signal characteristics in the DD2 band establishes a GA-BP model to predict the wear state of the pyramid sand belt with accuracy and stability.
查看全文  查看/发表评论  下载PDF阅读器
关闭

关于我们 | 联系我们 | 投诉建议 | 隐私保护 | 用户协议

您是第23386540位访问者    渝ICP备15012534号-3

版权所有:《表面技术》编辑部 2014 surface-techj.com, All Rights Reserved

邮编:400039 电话:023-68792193传真:023-68792396 Email: bmjs@surface-techj.com

渝公网安备 50010702501715号