航空学报 > 2022, Vol. 43 Issue (9): 625445-625445   doi: 10.7527/S1000-6893.2021.25445

基于重加权谱峭度方法的航空发动机故障诊断

张忠强1, 张新1, 王家序1, 刘治汶2   

  1. 1. 西南交通大学 机械工程学院, 成都 610031;
    2. 电子科技大学 自动化工程学院, 成都 611731
  • 收稿日期:2021-03-04 修回日期:2021-03-17 出版日期:2022-09-15 发布日期:2021-05-20
  • 通讯作者: 张新,E-mail:xylon.zhang@swjtu.edu.cn E-mail:xylon.zhang@swjtu.edu.cn
  • 基金资助:
    国家自然科学基金(52075456, 52075080);中央高校基本科研业务费专项资金(2682021CX021)

Reweighted kurtogram for aero-engine fault diagnosis

ZHANG Zhongqiang1, ZHANG Xin1, WANG Jiaxu1, LIU Zhiwen2   

  1. 1. School of Mechanical Engineering, Southwest Jiaotong University, Chengdu 610031, China;
    2. School of Automation Engineering, University of Electronic Science and Technology of China, Chengdu 611731, China
  • Received:2021-03-04 Revised:2021-03-17 Online:2022-09-15 Published:2021-05-20
  • Supported by:
    National Natural Science Foundation of China (52075456, 52075080); Fundamental Research Funds for the Central Universities (2682021CX021)

摘要: 针对快速谱峭度方法在分析含强冲击干扰信号时无法选取有效滤波器参数的问题, 定义了一种新的滤波器参数选择指标——重加权峭度, 并基于此提出了重加权谱峭度方法。该方法对经“Binary-Ternary”小波包分解后的各频段信号进行等分并计算各等分频段的峭度及所占其和的权重, 然后对各峭度和权重进行重排序并对峭度进行重加权得到重加权峭度, 最后将新指标表示在中心频率和带宽平面上得到重加权峭度图, 选取滤波器参数。仿真信号分析结果显示, 在强冲击干扰下重加权谱峭度方法仍能选择有效滤波器参数, 提取到周期性故障冲击。通过在航空发动机附齿轮箱中轴承故障诊断中的应用以及与常见方法的对比分析, 进一步验证了重加权谱峭度方法的有效性与优势。

关键词: 航空发动机, 故障诊断, 重加权谱峭度, 重加权峭度, 快速谱峭度, 滤波器参数

Abstract: To address the problem that the fast Kurtogram method cannot select effective filter parameters when analyzing some real vibration signals with strong impulse interferences, a reweighted kurtogram is proposed, where a new indicator-reweighted kurtosis, is defined. Firstly, the frequency bands obtained by decomposition of the "Binary-Ternary" wavelet packet are equally split into several segments. The kurtosis of each segment and their corresponding weights in their sum are calculated. Then, the kurtosis and weights are sorted in the descending and ascending order, and the reweighted kurtosis is calculated using the reordered kurtosis and weights. Finally, the reweighted kurtogram is obtained by representing the reweighted kurtosis in the plane of central frequency and bandwidth frequency. The simulated signal analysis results show the method proposed is still effective in extracting periodic fault impulses, even when the signal contains strong impulse interferences. Application in fault diagnosis of the aero-engine and comparisons with conventional methods further verify the effectiveness and advantages of the method.

Key words: aero-engine, fault diagnosis, reweighted kurtogram, reweighted kurtosis, fast kurtogram, filter parameter

中图分类号: