Parameter-based RNN micro-interface inversion model for wet friction components morphology

Abstract

The interface morphology significantly impact the service life of wet clutches friction components in heavy tracked vehicle transmission systems. This paper designs a sliding test and utilizes a recurrent neural network (RNN) model to construct the three-dimensional morphology of the wet clutch friction interface under specific operating conditions. It also explores the relationship between these factors and clutch performance. The interface morphology characteristics are analyzed by the RNN inversion model to assess the effects of three working condition parameters including rotational speed, pressure, and sliding time. This research provides important primary data for engineering studies and applications aimed at optimizing the design of wet clutches and improving transmission system reliability.