针对齿轮的振动特点,设计了复合过完备时频字典,利用基追踪方法在匹配信号特征结构、直接提取特征信息方面的优势分析了齿轮箱的现场测试振动信号.根据基追踪分解结果,在时频联合域内提取了齿轮局部损伤的周期性冲击特征,识别了齿轮点蚀缺陷.与短时Fourier变换和小波变换等时频分析方法进行了对比分析,验证了基追踪检测齿轮损伤的有效性. According to the characteristics of gear vibration, a composite over-complete time-frequency dictionary is designed, and the vibration signal of gearbox is analyzed by using the basic tracking method in the aspect of matching the signal characteristic structure and directly extracting the characteristic information.According to the result of base-chasing decomposition, The periodic impact characteristics of the gear local damage were extracted and the gear pitting defect was identified in the joint domain of the frequency domain. Compared with the time-frequency analysis methods such as the short-time Fourier transform and the wavelet transform, the effectiveness of the foundation tracking gear detection was verified.