文章摘要
胡超,沈宝国,杨妍,谢中敏.AdaBoost算法组合的GABP诊断模型在轴承故障中的运用[J].机床与液压,2021,49(2):163-169.
HU Chao,SHEN Baoguo,YANG Yan,XIE Zhongmin.Application of GABP Diagnostic Model Based on AdaBoost Algorithms in Bearing Fault [J].Machine Tool & Hydraulics,2021,49(2):163-169
AdaBoost算法组合的GABP诊断模型在轴承故障中的运用
Application of GABP Diagnostic Model Based on AdaBoost Algorithms in Bearing Fault
  
DOI:10.3969/j.issn.1001-3881.2021.02.33
中文关键词: 滚动轴承  故障诊断  GABP-AdaBoost  因子分析  BP神经网络
英文关键词: Rolling bearing  Fault diagnosis  GABP-AdaBoost  Factor analysis  BP neural network
基金项目:江苏省自然科学基金项目(BK20180863);镇江市科技计划资助项目(NY2019017;GY2018029);2020年度院级课题(JATC20010103)
作者单位E-mail
胡超 江苏航空职业技术学院航空工程学院 HuChao157@163.com 
沈宝国 江苏航空职业技术学院航空工程学院江苏航空职业技术学院镇江市无人机应用创新重点实验室  
杨妍 江苏航空职业技术学院航空工程学院  
谢中敏 江苏航空职业技术学院航空工程学院  
摘要点击次数: 77
全文下载次数: 0
中文摘要:
      考虑到轴承故障难以诊断的问题,提出AdaBoost算法组合遗传算法优化的BP神经网络(GABP-AdaBoost)的诊断模型。利用遗传算法寻优能力对BP网络的权值与阈值进行优化,并用AdaBoost算法进行组合;采用UCI标准数据集对GABP-AdaBoost算法中的关键参数进行分析,并设置最优参数;用最小二乘法和指数平滑法消除轴承振动信号中的漂移和微弱噪声,并用因子分析法选择最优时域参数;使用GABP-AdaBoost算法对轴承故障样本进行诊断,并将GABP、BP、BP-AdaBoost作为对比算法。重复试验30次的结果表明:GABP-AdaBoost算法诊断效果达到90%以上但诊断时间较长; BP-AdaBoost算法诊断效果优于GABP且耗时较少;GABP-AdaBoost算法与BP-AdaBoost算法对重复诊断的波动敏感程度较低
英文摘要:
      A diagnosis model of BP neural network optimized by genetic algorithm (GA) and AdaBoost algorithm was proposed,and it was used to address the problem of rolling bearing fault diagnosis.The weights and thresholds of BP network were optimized by GA (GABP),and multiple GABP was combined with AdaBoost algorithm as diagnosis model (GABP-AdaBoost).The key parameters of GABP-AdaBoost were analyzed based on UCI standard data set.The drift and weak noise in the bearing vibration signal was eliminated by least square method and exponential smoothing method,and factor analysis method was used to select the optimal time domain parameters.Finally,GABP-AdaBoost algorithm was used to diagnose the bearing fault samples,and GABP,BP and BP-AdaBoost were used as comparison algorithms.The results of 30 repeated experiments show that the diagnostic effect of GABP-AdaBoost algorithm is more than 90%,but the diagnostic time is longer;the diagnostic effect of BP-AdaBoost algorithm is better than that of GABP and less time consuming;and the sensitivity of GABP-AdaBoost algorithm and BP-AdaBoost algorithm to the fluctuation of repeated diagnosis is lower
查看全文   查看/发表评论  下载PDF阅读器
关闭

分享按钮