Joint envelope and frequency order spectrum analysis based on iterative generalized demodulation for planetary gearbox fault diagnosis under nonstationary conditions
Joint envelope and frequency order spectrum analysis based on iterative generalized demodulation for planetary gearbox fault diagnosis under nonstationary conditions
复制标题
非平稳工况下基于迭代广义解调的联合包络和频率阶谱分析行星齿轮箱故障诊断
DOI：
10.1016/j.ymssp.2016.02.047
复制
发表时间：
2016-08-01
影响因子：
8.4
通讯作者：
Liang, Ming
中科院分区：
文献类型：
--
作者：
Feng, Zhipeng;Chen, Xiaowang;Liang, Ming
行星变速箱振动信号在非平稳条件下具有时变性和复杂的多分量特征，使得故障诊断的特征提取非常困难。阶谱分析是旋转机械非平稳信号分析的有效方法之一。阶次分析的主要思想是将时变的频率分量映射成恒定的频率分量。在这一思想的启发下，我们提出了一种新的阶谱分析方法，利用迭代广义解调的独特性质，将多分量信号的任意瞬时频率轨迹转换为时频平面上的恒频线。这种新方法完全基于算法，不需要转速表/编码器，因此易于实现。它不涉及大多数阶数跟踪方法通常需要的等角度重采样，因此没有抽取和/或内插误差。本文提出的阶次分析方法可以消除频率的时变影响，从而有效地揭示非平稳多分量信号的谐波阶次成分。然而，行星齿轮箱的振动信号也会导致复杂的边带阶次。因此，我们进一步提出分析幅度包络的阶谱。这将消除原始信号阶谱中的复杂边带阶次，从而大大简化和更可靠的齿轮特性频率识别过程。然而，与齿轮故障无关的齿轮和/或行星架旋转频率顺序可能仍然存在。为了避免这种频率阶数可能导致的误导结果，我们还建议对瞬时频率的阶谱进行分析。从理论上讲，出现在频率阶谱上的峰值直接对应于齿轮特征频率阶数，可以更清晰地提取齿轮故障特征。通过对行星变速箱仿真信号的分析，说明了该方法的有效性，并用变速条件下的实验行星变速箱数据进行了验证。分析结果表明，该方法能有效地提取非平稳条件下齿轮的分布故障和局部故障。(C)2016爱思唯尔有限公司。保留所有权利。
Planetary gearbox vibration signals under nonstationary conditions are characterized by time-varying nature and complex multi-components, making it very difficult to extract features for fault diagnosis. Order spectrum analysis is one of the effective approaches for nonstationary signal analysis of rotating machinery. The main idea of order analysis is to map the time-varying frequency components into constant ones. Inspired by this idea, we propose a new order spectrum analysis method to exploit the unique property of iterative generalized demodulation in converting arbitrary instantaneous frequency trajectories of multi-component signals into constant frequency lines on the time-frequency plane. This new method is completely algorithm-based and tachometer/encoder-free, thus easy to implement. It does not involve equi-angular resampling commonly required by most order tracking methods and is hence free from the decimation and/or interpolation error. The proposed order analysis method can eliminate the time-variation effect of frequency and thus can effectively reveal the harmonic order constituents of nonstationary multi component signals. However, the planetary gearbox vibration signals also lead to complex sideband orders. As such, we further propose to analyze the order spectrum of amplitude envelope. This will eliminate the complex sideband orders in the order spectrum of original signals, leading to a substantially simplified and more reliable gear characteristic frequency identification process. Nevertheless, the gear and/or planet carrier rotating frequency orders, which are irrelevant to gear fault, may still exist. To avoid possible misleading results due to such frequency orders, we also propose to analyze the order spectrum of instantaneous frequency. Theoretically, the peaks present in frequency order spectrum directly correspond to the gear characteristic frequency orders, which can be used to extract gear fault signature more explicitly. The proposed method has been illustrated via analysis of the simulated signals of planetary gearboxes, and validated using lab experimental planetary gearbox datasets, both under variable speed conditions. The analysis results have shown that the method is effective in extracting both distributed and localized gear faults under nonstationary conditions. (C) 2016 Elsevier Ltd. All rights reserved.