A micromechanics-based machine learning model for evaluating the microstructure-dependent rolling contact fatigue performance of a martensitic steel

•The work determined the feasibility of using a low-cost machine learning model to predict fatigue performance of martensitic steel, and provided the micro-mechanistic basis of correlations between microstructure features and rolling contact fatigue.•The hierarchical microstructure modelling approach and size-dependent physics-based crystal plasticity framework were developed to capture the micromechanical response of the martensitic steel in rolling contact fatigue.•The hierarchical microstructure geometry, crystallographic orientation, and the corresponding fatigue indicator parameter obtained from crystal plasticity simulations were selected for database construction.•The initial database was filtered and order-reduced by considering the fatigue induced stress-strain response.