Affiliations: College of Engineering, Nanjing Agricultural University, Nanjing, Jiangsu 210031, China
Correspondence:
[*]
Corresponding author: Maohua Xiao, College of Engineering, Nanjing Agricultural University, Nanjing, Jiangsu 210031, China. E-mail: xiaomaohua@njau.edu.cn.
Abstract: Gearbox is one of the most important transmission equipment in mechanical equipment. The working status of gearbox has great influence on the whole machine and even the entire assembly line. However, the gearbox structure is precise, the matching precision is high, and the operating environment is harsh, so the frequency of failures is high. This paper takes a single-stage gearbox as an example to set three working conditions: normal, broken tooth and wear and tear, and collects corresponding vibration signals. It has explored the application of BP neural network, particle swarm algorithm and other technologies in gearbox fault diagnosis. Using the global search ability of the particle swarm algorithm to constantly search for the best weights and thresholds, and then give it to the BP neural network, and finally train the BP neural network optimized by particle swarm optimization. The PSO-BP algorithm proves its superiority in fault diagnosis.
Keywords: Gearbox, fault diagnosis, particle swarm optimization, BP neural network
