264 / 2025-06-15 18:52:35

Hole Expansion Life Prediction of Gear Interference-Fit in Locomotive Transmission Systems Based on GAN-BP Hybrid Model

interference fit,Plastic Hole Expansion,Damage Mechanism Model,Neural Network,Lifetime Prediction System

In locomotive transmission systems, interference fit between gears and shafts is prone to plastic hole expansion due to the coupling effects of dynamic loading and residual stresses. This article systematically investigates the coupled influence of interference magnitude, residual stresses and dynamic load on the stress distribution within gear hole. A damage mechanism model for plastic hole expansion is established based on material damage mechanics. Stress data are acquired through finite element simulation and a hybrid prediction model integrating Generative Adversarial Networks (GAN) with Backpropagation (BP) Neural Networks is proposed to predicting the maximum Mises stress of the gear hole. A lifetime prediction system is developed by coupling this hybrid model with the damage mechanism model. The results show that the interference range (0.241–0.3 mm) of the gear design satisfies strength requirement, with the maximum relative error between predicted plastic hole expansion values of  the plastic hole expansion prediction system and measured plastic hole expansion values is 5.4%. It effectively quantifies the evolution of service-induced damage. The research provides a robust theoretical and technical foundation for reliability assessment of transmission systems.