计算机工程与应用 ›› 2019, Vol. 55 ›› Issue (16): 123-129.DOI: 10.3778/j.issn.1002-8331.1903-0010

• 模式识别与人工智能 • 上一篇    下一篇

基于深度神经网络的提升机轴承故障诊断研究

马辉,车迪,牛强,夏士雄   

  1. 中国矿业大学 计算机科学与技术学院,江苏 徐州 221116
  • 出版日期:2019-08-15 发布日期:2019-08-13

Research on Fault Diagnosis of Hoisting Bearing Based on Deep Neural Network

MA Hui, CHE Di, NIU Qiang, XIA Shixiong   

  1. School of Computer Science and Technology, China University of Mining and Technology, Xuzhou, Jiangsu 221116, China
  • Online:2019-08-15 Published:2019-08-13

摘要: 为了对矿井提升机的轴承故障进行精确诊断,提出一种基于深度神经网络的双层次故障诊断系统,精准识别提升机轴承的故障类型及故障程度。该系统首先利用滑动窗口重叠采样技术进行数据增强,随后结合自编码器减少噪声影响,通过反向传播算法训练深度神经网络双层分类器识别出故障模式及故障程度,最后用集成学习投票法进一步提高识别准确率。实验结果表明,该系统诊断准确率高于SVM与BPNN算法,可以完成提升机轴承的故障诊断任务。

关键词: 提升机轴承, 神经网络, 故障诊断, 自编码, 分类

Abstract: In order to diagnose the fault of hoisting bearings for the mine system. This paper proposes a multi-level diagnosis system based on deep learning to dynamically identify the type of fault and the degree of hoisting bearing degradation. Firstly, the sliding window overlapping sampling is used for realizing data enhancement in the proposed system, then the data noise is reduced based on auto-encoder. Through the training of deep neural networks, the failure mode and the degree of failure are identified by classification. Finally, in order to further improve the accuracy, this paper adopts the voting method of integrated learning. The experimental results show that the accuracy of system is higher than SVM and BPNN algorithms. Therefore, the system can identify the type and degree of hoisting bearing fault.

Key words: hoisting bearing, neural network, fault diagnosis, auto-encoder, classification