# Application of wavelet analysis in fault diagnosis of tractor gearbox

The tractor gearbox is a complex device in the basic components of the whole machine, and the failure rate of the gearbox accounts for 50% – 70% of the failure rate except for the engine. Therefore, it is of great significance to study and explore the mechanism, mode and diagnosis method of tractor gearbox fault for the use and maintenance of tractor.

In the past, when analyzing and processing the fault vibration signal of gearbox, Fourier analysis (i.e. spectrum analysis) was usually used to analyze and process the measured signal. But in the actual work, the impact and friction caused by the unstable rotating speed, load change and machine failure will lead to the generation of non-stationary vibration signals. If we continue to use the spectrum analysis method to analyze these abnormal signals, we may get false results or no diagnosis results, which will affect the accuracy of the diagnosis results. Therefore, in the process of fault diagnosis and signal monitoring, time-frequency analysis should be carried out for the signal, so as to have a deeper analysis and understanding of the signal. The time-frequency analysis of the signal requires a mathematical method, which can localize the signal in the time domain and frequency domain at the same time. In recent years, wavelet analysis, which appears and can be applied and developed, is a time-frequency signal analysis method proposed to overcome the limitations of traditional Fourier analysis and convolution filter analysis. Because of its localization in time and frequency domain and the characteristics of variable time-frequency window, it has more significant advantages than the traditional Fourier analysis. However, the fault vibration signal of tractor gearbox is very complex, which also contains unsteady components, so it is feasible to use wavelet analysis to deal with the vibration signal.

Proper selection of the number of decomposition layers can obtain the required frequency band width and the start and stop frequencies of each frequency band, so that the useful components, interference and noise in the original signal can be separated. At the same time, because each frequency band has a certain width, the frequency of the components to be retained in the original signal does not need to be accurately located, so it has a certain adaptability to frequency drift.

By using the reconstruction of wavelet packet, we can select all or sub frequency information according to the needs, and do not clear the rest frequency (interference noise) to reconstruct the signal. Therefore, as long as the useful components of the signal, noise and interference can be decomposed into different channels, the useful components of the original signal can be easily reconstructed.

It can be seen from the above that, because of the above characteristics of wavelet packet, it is very helpful to extract the useful components of signal in the wide-band range, and has certain adaptability to frequency drift. Therefore, it can be applied to the extraction of impact response features and is convenient for automatic fault diagnosis.

This paper focuses on the working process of fault diagnosis of gearbox by using continuous wavelet transform. It is considered that continuous wavelet transform is very effective to accurately identify impact signal and has strong anti noise ability, especially suitable for detecting weak impact signal. The vibration acceleration data of the gearbox of Dongfeng 81-a walking tractor show that the continuous wavelet transform based on Morlet wavelet basis can effectively identify the weak fault of the gearbox teeth of the tractor, which is superior to the traditional Fourier analysis fault diagnosis method. Wavelet packet analysis technology is very helpful to extract the useful components of signal in the wide frequency band, and has certain adaptability to frequency drift, which provides a way to extract fault features for the establishment of automatic fault diagnosis system.

Scroll to Top