As industrialization continues to progress, the level of development in machinery and equipment has emerged as a critical metric for assessing a country's core competitiveness. As a key equipment, gear transmission devices are widely used in energy, petrochemical, mining, automotive and other fields, and increasingly tend to be large-scale, complicated and intelligent. If you can always know the health of the gearbox and quickly and accurately identify the fault category, you can improve the safety of the gear drive. Based on the herringbone gear data of a paralleling gear box of a certain institute, this study carried out mechanical feature extraction for gear faults through Fourier transform, Hilbert-Huang transform, Short-Time Fourier transform and Continuous Wavelet transform, and verified the actual gear rotation parameters such as rotation frequency, meshing frequency and spectral diagram, envelope spectrum diagram and time spectrum diagram. Ensure the rationality and effectiveness of the used methods.