Research on high-speed drag torque characteristics of wet clutches based on mechanism and data-driven approach

The traditional drag torque model can accurately predict the drag torque in the low rotation speed stage, but cannot predict the drag torque rebound change in the high rotation speed stage. Therefore, a hybrid model based on the traditional wet clutch drag torque model and Particle Swarm Optimization-Back Propagation (PS0-BP) neural network is proposed in this paper, and the accuracy of the model is improved by the test data. The results show that the error of this hybrid model is 14.45%, which is better than the traditional drag torque model, and the stability and reliability are significantly improved compared with the other neural network models. The effects of oil temperature, the clearance of the friction pair, and the flow rate of lubricant on the drag torque are investigated. It was found that, with the increase of oil temperature and clearance of the friction pair, the rotational speed corresponding to the rebound change of drag torque decreases, and drag torque decreases. With the increase of the flow rate of lubricant, the rotational speed corresponding to the rebound change of drag torque rises, and drag torque increases.