摘要
为提高行星齿轮箱健康评估准确性,提出一种基于图谱特征与度量学习的行星齿轮箱健康评估方法。从行星齿轮箱振动信号中提取图谱特征作为故障特征参数;设计基于单调性、相关性的度量学习准则,建立优化的马氏距离度量函数;采用待测样本与无故障正常样本之间的马氏距离表征故障严重程度,建立基于支持向量回归的健康评估模型。通过行星齿轮箱健康评估实验结果分析,证明了图谱特征能够有效表征行星齿轮箱故障严重程度,所建立的健康评估模型单调性好,提高了健康评估准确性。
To improve the accuracy of planetary gearbox health assessment, a method for planetary gearbox health assessment based on spectral graph features and metric learning is proposed. The spectral graph features is extracted from the vibration signal of the planetary gearbox as the fault feature parameter.The metric learning model based on monotonicity and correlation is design to establish the optimized Mahalanobis distance function. The severity of characterization fault of Mahalanobis distance between the sample to be tested and the normal sample is used to build a health assessment model based on support vector regression. The analyzed results of the planetary gearbox health assessment experiment prove that spectral graph features can effectively characterize the severity of the planetary gearbox failure, and the established health assessment model has good monotonicity, which improves the accuracy of the health assessment.
作者
李加兴
陈广艳
张鲁晋
王友仁
张砦
LI Jiaxing;CHEN Guangyan;ZHANG Lujin;WANG Youren;ZHANG Zhai(College of Automation Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China;Key Laboratory of National Defense Science and Technology of Helicopter Transmission Technology,Hunan Power Machinery Research Institute of Aviation Development,Zhuzhou 412000,China)
出处
《机械制造与自动化》
2022年第2期107-110,119,共5页
Machine Building & Automation
基金
直升机传动技术国防科技重点实验室基金项目(KY-52-2018-0024)
航空科学基金项目(20183352031)。
关键词
行星齿轮箱
健康评估
马氏距离
图谱特征
度量学习准则
支持向量回归
planetary gearbox
health assessment
Mahalanobis distance
spectral graph feature
metric learning model
support vector regression
作者简介
第一作者:李加兴(1996—),男,山东枣庄人,硕士研究生,研究方向为航空设备故障诊断与健康评估。
