Experimental study on mechanical properties of thick rubber bearings and vertical stiffness prediction using Sparrow search algorithm-back propagation neural network (SSA-BP) model

Abstract

Thick-layer rubber bearings exhibit a more pronounced steel plate constraint effect due to the increased rubber layer thickness. However, current standards only specify the steel plate thickness range for ordinary rubber bearings, without providing dedicated guidelines for thick-layer rubber bearings. Through experimental research and simulation analysis, this study investigates the impact of steel plate thickness on the mechanical performance of thick-layer rubber bearings, refines the relevant theoretical formulas, and develops a predictive model based on the SSA-BP neural network. A thick-layer rubber bearing with a first shape factor of 2.84 was designed, and vertical compression and horizontal shear tests were conducted, complemented by refined numerical simulations. The results reveal that steel plate thickness significantly affects vertical stiffness but has a minimal impact on horizontal stiffness. When the steel plate thickness is 6 mm (thickness ratio 0.4), the vertical stiffness reaches over 80% of its maximum value, while increasing the thickness to 14 mm (thickness ratio 0.93) reduces the change in vertical stiffness to within 5%. Based on upper and lower bound analyses of ultimate stress and yield plasticity ratios, the study recommends that the steel plate thickness should be 1.3 to 1.6 times the rubber thickness, which is a key parameter for optimizing the performance of thick-layer rubber bearings. The developed SSA-BP neural network prediction model for the vertical stiffness of thick-layer rubber bearings demonstrates excellent generalization capability and high predictive accuracy.