General multi-tooth contact analysis of spiral bevel and hypoid gears with arbitrary shaft angles considering the point clouds reconstruction of gear surface based on deep neural network

Abstract

Tooth contact analysis is one of the most significant means to support the high-quality spiral bevel and hypoid gears design. This paper proposes a general multi-tooth contact analysis methodology of spiral bevel and hypoid gears with arbitrary shaft angle configurations considering the point clouds reconstruction of the gear surface. Crossing point sets of tooth flanks are established by transforming coordinates into a global coordinate system, enabling pre-contact calculations. The pre-contact calculation of the rotation angle has been derived to provide the basis for an accurate contact path determination. When the minimum distance falls below a specified tolerance, the potential contact region is identified, centered on the point of the global minimum distance. A deep neural network-based surface fitting algorithm generates a polynomial representation of the potential contact region. Based on the contact ratio and contact distance region, the contact pattern considering multi-tooth alternating contact can be performed. Finite element analysis and experimental validation confirm strong agreement, demonstrating the method's efficient.