基于非线性短时傅里叶变换阶次跟踪的变速行星齿轮箱故障诊断

针对变速行星齿轮箱信号频率模糊且受噪声影响的问题，提出了基于非线性短时傅里叶变换（NLSTFT）无键相阶次跟踪与变分模态分解的故障诊断方法。用NLSTFT算法估计信号瞬时频率，对其积分获得瞬时相位曲线，通过重采样得到角域信号；利用NCOGS算法对角域信号降噪，采用VMD算法进行角域信号模态分解，通过各模态分量信号包络谱解调实现故障诊断。实验结果表明，新方法计算效率高、鲁棒性好，提高了变转速行星齿轮箱故障诊断性能。

Fault Diagnosis of Planetary Gearboxes Based on NLSTFT Order Tracking under Variable Speed Conditions

Aiming at the problems that vibration signal frequencies of planetary gearboxs were fuzzy and interfered by strong noises under variable speed conditions，a fault diagnosis method was proposed based on NLSTFT non-bonding phase order tracking and variational modal decomposition. Firstly，NLSTFT was used to precisely estimate instantaneous frequency of vibration signals，and then transformed into the instantaneous phase curves through integral operation. A smooth angular domain signal was obtained by signal resampling. Secondly，NCOGS（non-convex overlapping group shrinkage） algorithm was used to de-noise the angular domain signals，and VMD（variational mode decomposition） algorithm was used to perform modal decomposition of the angular domain signals. Finally，the fault diagnosis was realized through the demodulation analysis of the envelope spectrum of each modal component signal. The experimental results show that the new method may effectively promote the performances of fault diagnosis under variable speed conditions，which has the high computational efficiency and strong noise robustness.