Multi-Objective Optimization of Active Hybrid Fluid Film Bearings Using Heuristic Algorithms

Abstract

Purpose of research. The design of sliding bearings, especially for heavily loaded rotary machines, is a laborious task. The implementation of control systems for the movement parameters of the rotor further increases the complexity of a design procedure. The study shows a developed approach to the optimal design of active rotor bearings using heuristic optimization algorithms. The approach allows to obtain a set of optimal Pareto solutions and determine the only configuration of the reference node that best meets the given criteria.Methods. The problem of optimal parametric synthesis of an active fluid friction bearing was solved using a numerical model coupled with the model of rotor movement in the support. For the given design problem, objective functions were formulated, design variables were determined, and the necessary restrictions were imposed. Using multicriteria versions of the genetic algorithm and the particle swarm algorithm, procedures for the optimal synthesis of reference nodes were carried out. The solutions obtained by different methods are compared and analyzed based on the results of model tests.Results. As part of the study, algorithmic and software tools were developed for solving problems of optimal parametric synthesis of active hybrid fluid friction bearings. The applied objective functions are conflicting, so the primary result of the solution is a 3D Pareto front. The tested heuristic algorithms showed qualitatively similar solutions, but the genetic algorithm covers a larger range of them. On the whole, the final decisions meet the criteria, but the methods for making final decisions require additional elaboration.Conclusion. The study presents an approach to the automated design of sliding bearings, which allows you to simultaneously take into account the tribological, dynamic aspects of the behavior of the rotary bearing system, as well as ensure readiness for the use of control systems in bearing nodes. The tested heuristic algorithms give comparable solutions to the optimization problem in comparable time as well. Further improvement of the method of parametric synthesis of such supports should be carried out in the direction of decision-making algorithms, refinement of objective functions, as well as acceleration of the applied calculation models.