涡桨发动机减速器轴承的局部故障会严重影响发动机的运行安全。由于减速器结构复杂、运转高速，导致作为监测机械零件运行状态的振动信号成份十分复杂。为了能够从振动信号中实时检测到轴承故障信号成份和准确地提取轴承的故障特征，提出了基于振动信号的快速傅里叶变换（fast Fourier transform, FFT）频谱、快速谱峭度图（fast kurtogram）算法、包络谱（envelop spectrum）相结合的故障检测和诊断方法。首先，在发动机运行过程中使用频谱分析检测振动信号中是否存在轴承故障冲击信号成份，然后采用快速谱峭度图确定故障信号的分布频带。最后，运用包络谱分析提取故障特征频率，根据故障特征频率确定故障发生的具体部位。在某型号涡桨发动机地面台架试验过程中，使用此方法准确检测和诊断了减速器轴承内滚道的剥落故障。因此，此方法可以为航空发动机减速器轴承故障检测与诊断技术的工程应用提供依据。

Bearing local fault occurs in the turboprop engine reduction gearbox, which affects engine running safely. Due to the complex structure and high speed of the reducer gearbox, the vibration signal used to monitor the operation state of the mechanical components is complicated. In order to detect the bearing local fault signal in time and extract fault feature accurately from the vibration signal, a method which was based on vibration signal combining FFT, fast kurtogram and envelope spectrum was proposed. Firstly, during the engine operation, the FFT(fast Fourier transform) spectrum was used to detect whether there was bearing fault component in the vibration signal; and then the fast kurtogram was applied to determine the frequency band distribution of the fault component; finally, the bearing fault characteristic frequency was acquired through envelope spectrum analysis. In the course of certain type of turboprop engine ground bench test, this method was used to accurately detect and diagnose the local spalling fault of the inner raceway. Therefore, this method can provide a basis for the engineering application on local fault detection and diagnosis of bearings in aero-engine reduction gearbox.