Triboelectric sensor for planetary gear fault diagnosis using data enhancement and CNN

Abstract

The planetary gear transmission system is characterized by a complex structure and is often exposed to harsh working conditions, resulting in high failure damage rate. The self-powered sensing system based on the triboelectric nanogenerator provides an ideal scheme for online monitoring of the planetary gear operating condition. In this paper, a triboelectric planetary gear sensor (TPGS) is proposed for real-time monitoring of the speed and gear faults of the planetary reducer. The sensor is integrated at the tooth crest and root meshing gap of the planetary gear and the ring gear, which ensures the integrity of the gear structure and function. The results of the sensing experiment show that TPGS can be used to detect the speed of the planetary reducer in the range of 10–1000 rpm. Moreover, the speed detection error rate is less than 0.4%. To identify the gear faults, the stacked autoencoder (SAE) is used to enhance the data collected by TPGS, combined with convolutional neural network (CNN). The results show that the classification accuracy reached 100%. In this paper, a theoretical and experimental basis for development and online fault monitoring of a new intelligent gearbox with self-awareness and self-regulation is provided.