Because planetary gearbox has larger transmission ratio and higher load efficiency than traditional fixed shaft gearbox, it is widely used in aerospace, wind turbine and lifting machinery. However, the strong background noise brought by the complex working environment will submerge part of the weak fault features, so the research on the extraction method of weak fault features of planetary gearbox has become the focus and difficulty in the research of planetary gearbox fault diagnosis.
At present, researchers have used many methods to study the fault characteristics of planetary gearbox. In this paper, a feature extraction method based on deep learning diversity is proposed. The method combines with information fusion technology, which effectively improves the fault diagnosis accuracy and stability of planetary gearbox. An intelligent fault diagnosis method based on empirical mode decomposition (EMD) and deep convolution neural network (DCNN) is proposed, which can accurately and effectively identify the working state and fault type of planetary gearbox. The dual tree complex wavelet is introduced into the vibration signal analysis to realize the operation monitoring and fault type identification of wind turbine.
At present, many fault diagnosis methods of planetary gearbox are only transferred from the traditional fault diagnosis method of fixed shaft gearbox. Therefore, many scholars have carried out relevant research on the complex vibration signal transmission path and the existence of multiple characteristic frequencies in planetary gearbox. Lei Yaguo et al. Established the dynamic model of planetary gear system, and summarized the vibration signal characteristics of planetary gear box partial fault combined with the test; Feng Zhipeng et al established the vibration signal model of normal and fault of planetary gear box, and used envelope spectrum, instantaneous Fourier spectrum and other methods to diagnose the fault of planetary gear box; Nejad ar et al. Studied the operating characteristics of star gear box under fault condition.
It can be seen from the above: when the planetary gearbox fails, there are many frequencies in the vibration signal, such as the meshing frequency, the rotating frequency of each component and the fault characteristic frequency of each component, and these frequencies will be coupled and modulated in various forms, which makes the fault characteristics more complex. Therefore, it is necessary to decouple the various frequency coupling in planetary gearbox.
By applying 1.5-dimensional spectrum to complex domain and combining with envelope analysis, the coupling of planetary gearbox in different wear stages is obtained, and the 1.5-dimensional envelope spectrum is generated. The frequency coupling law of planetary gearbox wear fault is extracted and summarized, which provides a useful basis for fault diagnosis of planetary gear box.