Optimization tool based on multi-objective adaptive surrogate modeling for surface texture design of slipper bearing in axial piston pump

Abstract

A novel multiobjective adaptive surrogate model is employed for the prediction of full-film lubrication performance of surface texture profile for thrust bearings in axial piston pumps. At present, a surface texture optimization design of thrust bearing in axial piston pump has not been provided based on a unique design solutions. In this study, multi-objective adaptive surrogate-based optimization (MO-ASMO) algorithm was used for solving both the mechanical and volumetric losses performance design of surface texture. The non-dominated sorting genetic algorithm (NSGA-II) formed part of the developed multi-objective optimization model, having three competing design objectives which mainly include load capacity, friction torque and leakage rate. A comparative study of two models with accuracy analysis of each case was performed. The results show that the MO-ASMO has the best performance among the different surrogate models. Comparing with the NSGA-II model, the improvement of the leakage rate and the load carrying force made by MO-ASMO are higher about 16.12% and 3.67%, respectively. In other words, multi-objective optimization is capable of enhancing textured slipper bearing performance using MO-ASMO method. The optimal texture radius and texture depth are chosen to minimize the leakage rate and texture bearing capacity.