Dynamic tension prediction in belt drive assembly and its application

LIN Xiao-han, WANG Shao-jie, HOU Liang, YANG Zheng, MU Rui
（Department of Mechanical and Electrical Engineering, Xiamen University, Xiamen 361005, China）

Abstract: Aiming at the problems of low accuracy in the measurement and inaccuracy adjustment of tension in the assembly, a tension measurement and a prediction method based on neural network and vibration signal analysis were proposed. Firstly, the steady-state tension measurement and vibration signal acquisition test were designed to achieve convenient and accurate condition measurement and signal acquisition. Secondly, the vibration signals under different tension were obtained through the bench simulation test of belt drive, and the tension prediction research based on BP, RBF and GRNN neural networks was carried out to achieve accurate tension prediction. Finally, the above test method and prediction network model were applied to the tension prediction of a belt drive. The results indicate that GRNN model is fast and accurate, and the probability of bus tension prediction error in a reasonable range is 86.15%. The proposed method is an effective way to achieve tension prediction, which is of great significance to optimize assembly and improve product performance.

Key words: belt drive; belt tension; assembly quality; recognition and prediction; neural network