Bi-level optimization and integrated multiple-criteria decision making under uncertainty for structural design of insulated bearings

Abstract

To address the premature failure of insulated bearings, this article proposes a design method for the structure of bearings to prolong their service life. First, a bi-level optimization model is established to determine the structural parameters of bearings considering their insulating and mechanical properties. Upper-level problems include bearing voltage ratio, porosity of insulating coatings, minimum thickness of oil film and Hertz contact area. Lower-level problems include fatigue life of bearings and adhesion of insulating coatings. Second, a non-dominated sorting genetic algorithm-III based on space decomposition and penalty-based boundary intersection (NSGA-III-SD-PBI) is designed to solve the optimization model. Third, the integrated multiple-criteria decision-making (MCDM) using the fuzzy best–worst method (BWM) and fuzzy technique for order preference by similarity to an ideal solution (TOPSIS) is developed to determine the best structural parameters of bearings. A case study demonstrates the effectiveness and superiority of the proposed method, with fatigue life increased by 20%.