航空学报 > 2018, Vol. 39 Issue (9): 422025-422036   doi: 10.7527/S1000-6893.2018.22025

基于IGA-ELM网络的滚动轴承故障诊断

皮骏1, 马圣2, 贺嘉诚2, 孔庆国3, 林家泉4, 刘光才1   

  1. 1. 中国民航大学 通用航空学院, 天津 300300;
    2. 中国民航大学 航空工程学院, 天津 300300;
    3. 中国民航大学 中欧航空工程师学院, 天津 300300;
    4. 中国民航大学 电子信息与自动化学院, 天津 300300
  • 收稿日期:2018-01-17 修回日期:2018-03-14 出版日期:2018-09-15 发布日期:2018-06-01
  • 通讯作者: 皮骏 E-mail:pijun8@163.com
  • 基金资助:
    中央高校基本科研业务费项目中国民航大学专项资助(3122017056)

Rolling bearing fault diagnosis based on IGA-ELM network

PI Jun1, MA Sheng2, HE Jiacheng2, KONG Qingguo3, LIN Jiaquan4, LIU Guangcai1   

  1. 1. General Aviation College, Civil Aviation University of China, Tianjin 300300, China;
    2. College of Aeronautical Engineering, Civil Aviation University of China, Tianjin 300300, China;
    3. Sino-European Institute of Aviation Engineering, Civil Aviation University of China, Tianjin 300300, China;
    4. College of Electronic information and Automation, Civil Aviation University of China, Tianjin 300300, China
  • Received:2018-01-17 Revised:2018-03-14 Online:2018-09-15 Published:2018-06-01
  • Supported by:
    Central University Basic Research Business Fee Project Civil Aviation University of China Special Funding (3122017056)

摘要: 为了提高航空发动机轴承故障诊断准确率,提出基于改进遗传算法优化极限学习机网络(IGA-ELM)的诊断模型。针对传统遗传算法易早熟等缺陷,对遗传算法的交叉操作和变异操作进行改进,并用改进的遗传算法优化极限学习机的输入权值矩阵和隐含层阈值,利用Moore-Penrose算法计算极限学习机的输出权值矩阵。使用IGA-ELM诊断模型对滚动轴承正常、内环故障、外环故障和滚珠故障4种工况进行诊断,并分析极限学习机隐含层神经元的数量和激活函数对轴承故障诊断的影响。为了验证改进遗传算法优化极限学习机的有效性,将传统遗传算法、自适应遗传算法和粒子群算法作为对比算法。经过分析表明:改进遗传算法收敛速度和收敛误差,均优于对比算法。

关键词: 航空发动机, 轴承故障诊断, 极限学习机, 交叉操作, 变异操作, 遗传算法

Abstract: To improve the accuracy of fault diagnosis of aero-engine bearings, a diagnosis model is proposed based on the Optimized Extreme Learning Machine by Improved Genetic Algorithm (IGA-ELM). Crossover operation and mutation operation of the genetic algorithm is improved in this paper to overcome the flaw of the traditional genetic algorithm. The input weight matrix of the connection between the input and hidden layers and biases of hidden neurons of the extreme learning machine is optimized by the improved genetic algorithm. The output matrix of the connection between the hidden and output layers was calculated by using the Moore-Penrose algorithm. Four cases including normal rolling bearing, inner ring fault, out ring fault, and ball fault are diagnosed by using the optimized extreme learning machine. The influence of the number of the hidden layer neurons and the activation function on bearing fault diagnosis of the extreme learning machine is also analyzed. To verify to, as a comparative algorithm A comparison of the traditional genetic algorithm, adaptive genetic algorithm and particle swarm optimization algorithm shows the effectiveness of the improved genetic algorithm in extreme learning machine optimization. Analysis shows that the improved genetic algorithm are better than the other algorithms in terms of convergence speed and convergence error.

Key words: aero-engine, bearing fault diagnosis, extreme learning machine, crossover operation, mutation operation, genetic algorithm

中图分类号: