Abstract
This article aims to provide an efficient fault diagnosis method for gearbox. A self-organizing map–based fault model is developed to provide effective diagnosis of the faults of gearboxes using the gear signals extracted from gearboxes operating with zero and three different types of faults. The gear signals are collected by vibration and acoustic sensors, and pre-denoised using wavelet denoising and wavelet packet decomposition. The characteristic values are subsequently obtained using fast Fourier transform and infinite impulse response filtering. The results showed of the self-organizing map neural network diagnosis model can effectively diagnose gear fault information with a 95% diagnostic accuracy using four input characteristic values: (1) Y-axis vibration displacement amplitude, (2) Y-axis vibration acceleration amplitude, (3) acoustic emission energy amplitude, and (4) acoustic emission signal peak value. The proposed approach provides a novel method to more accurate diagnosis of gear fault pattern and improvement of working efficiency of mechanical instruments.
Highlights
Gearboxes have been widely used in modern mechanical instruments as the key transmission components for change of speed and transmission of power
Liu et al.[5] proposed a characteristic value extraction method of gear fault based on the envelope analysis and time– frequency image of S transformation
Yang et al.[6] proposed a method of gear fault diagnosis based on multiscale fuzzy entropy of ensemble empirical mode decomposition (EEMD)
Summary
Gearboxes have been widely used in modern mechanical instruments as the key transmission components for change of speed and transmission of power. Liu et al.[14] proposed a feature extraction and fault diagnosis method based on the variational mode decomposition, SVD, and convolutional neural network for the local weak feature information of planetary gears. Some studies have explored gear fault diagnosis using a variety of signal fusion methods,[15,16] but the denoising effect of the signal is not obvious, resulting in inaccurate extraction of eigenvalues. Such an approach suffers from the drawback of low accuracy of the characteristic value extraction and long operating time due to the absence of signal fusion and poor performance of denoising, so there will be inefficient diagnosis, low diagnostic rate, and other issues.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: International Journal of Distributed Sensor Networks
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
