Investigation of loaded contact characteristics of planetary roller screw mechanism based on influence coefficient method and machine learning

Abstract

The performance of the planetary roller screw mechanism (PRSM) is directly determined by the mechanical behavior of the loaded contact. The traditional method based on a finite element modeling to describe the loaded contact characteristics (LCCs) is computationally expensive, while the use of Hertzian contact theory and the thread connection model results in a loss of computational accuracy. As a solution, this study aims to develop a semi-analytic modeling technique that can predict the LCCs of PRSM accurately and efficiently. To achieve this, the study first establishes a mathematical model using the spatial coordinate transformation theory to accurately describe the thread surface. The potential contact area is discretely divided using the differential geometry principle. Then, the linear deformation and nonlinear deformation are matched using node interpolation technique to obtain the accurate compliance of the potential contact area. Finally, the contact forces for all activated contact points are derived by a nonlinear iterative algorithm based on the influence coefficient method. The performance of the proposed model is verified by numerical examples. Further, the sensitivity weights of the geometric parameters to the LCCs are identified utilizing machine learning, and a preferred zone is also constructed to optimize these parameters. This study provides a valuable tool for evaluating and improving the LCCs of PRSM in engineering practice.