Fault diagnosis of aviation bearing based on enhanced multi fractional domain feature

Due to the variable speed of the aircraft during flight, the collected signal components are diverse and the transmission path is complex. For the important aviation bearing, the features extracted by traditional methods are difficult to effectively reflect the health state of the bearing and will affect the diagnosis accuracy. Based on the traditional dimensionless index in time domain, this paper considers the multi-fractional domain feature extraction of variable speed bearing signal with fractional Fourier transform. At the same time, the feature enhancement method is used to improve the feature separability. The enhanced multi-fractional domain features realize effective bearing fault diagnosis by random forest classification model. This paper illustrates the advantage and feasibility of the proposed theory through the variable speed bearing data set. The analysis results show that the enhanced multi-fractional domain feature has better classification performance than the original time domain feature, and the classification effect of random forest in this algorithm is better than other classification models.