Dynamic modeling of defective bearing with spatially dependent contact force considering instantaneous position of the ball and three-dimensional defect area

Abstract The dimensions of localized surface defects, e.g. width, length, and depth, significantly influence the vibration characteristics of bearings. Understanding the fault mechanism necessitates a dynamic model capable of accurately simulating these characteristics. Its accuracy heavily relies on the precise portrayal of the interaction between the ball and the defect boundaries within the model. However, most existing studies use pre-planned excitations, so neglecting the actual interaction dynamics between the ball and defective raceway. This paper proposes an improved model for defective bearings that considers interactions via the instantaneous positions of the ball and the three-dimensional defect area. It approximates the spatially dependent force deflection relationship using an artificial neural network, which characterizes ball-line contact between the ball and defect edges. This specific relationship cannot be described by a closed-form solution derived from classical contact theory. The validity and accuracy of proposed model are confirmed by comparisons with experimental responses from defective deep groove ball bearings. This study offers a solid quantitative basis for investigating the underlying mechanisms responsible for abrupt variations in contact force and the associated vibration characteristics due to localized surface defects. Its correlation between actual defect dimensions and vibration measurement characteristics has the potential to advance diagnostic and prognostic algorithms, as well as data augmentation for deep learning techniques.