Gearbox fault detection commonly relies on comparing sideband frequencies beside the meshing frequency of gear, and its harmonics of higher orders in frequency domain between the measured signal and healthy baseline. However, this type of localized fault detection algorithm is less intuitive and credible because other factors, such as manufacturing and assembly errors, can lead to similar sidebands without a localized fault. As such, this study proposes a novel and direct gearbox-localized fault detection strategy by taking full advantage of the meshing frequency modulation characteristic. During application, a baseline frequency band that is modulated by the meshing frequency was first located via a signal decomposition algorithm and meshing frequency modulation (MFM) index under healthy conditions. The raw signal was then analyzed using the same process. If there is a localized fault in the monitored gearbox, the bandwidth of the baseline frequency band will increase. Subsequently, a new frequency band will emerge around the higher-frequency zone, which is modulated by the gear meshing frequency. Compared with a healthy condition, a gear fault introduces a new MFM area. The usefulness of the proposed algorithm was demonstrated using the synthetic signal and experimental signal from the testrig.

中文翻译：

---

基于啮合调频分析的齿轮箱局部故障检测

齿轮箱故障检测通常依赖于比较齿轮啮合频率之外的边带频率，以及测量信号和健康基线之间频域中的高阶谐波。然而，这种类型的局部故障检测算法不太直观和可信，因为其他因素（例如制造和装配误差）可能会导致类似的边带而没有局部故障。因此，本研究充分利用啮合调频特性，提出了一种新颖、直接的齿轮箱局部故障检测策略。在应用过程中，首先通过信号分解算法和健康条件下的啮合频率调制（MFM）指数来定位由啮合频率调制的基线频带。然后使用相同的过程分析原始信号。如果受监控的齿轮箱出现局部故障，基线频带的带宽将会增加。随后，在较高频率区域周围将出现一个新的频带，该频带由齿轮啮合频率调制。与健康状况相比，齿轮故障引入了新的 MFM 区域。使用来自测试台的合成信号和实验信号证明了所提出算法的实用性。