Multi-Objective Optimal Design of Rotor-Bearing Systems under Dynamic Behavior Constraints Using a Hybrid Genetic Algorithm

Abstract

A new hybrid optimization procedure of rotorbearing systems, which combines the genetic algorithm (GA) with traditional optimization methods, is presented in this paper. Most traditional optimization methods applied in engineering design require a better set of initial values for the design variables, and then converge rapidly to generate good results. In the first step of the procedure, a GA is applied to provide a set of initial design variables, thereby avoiding the trial process; thereafter, traditional algorithms are employed to determine the optimum results. This hybrid algorithm, which can be termed a hybrid genetic algorithm (HGA), is more effective than the traditional ones. The capacity of the HGA is demonstrated by the optimization of rotor-bearing systems under dynamic behavior constraints. The optimization involves minimizing, either individually or simultaneously, the shaft weight and the transmitted forces at the bearings. The results show that an HGA can identify more effectively better initial design variables. Moreover, it can identify superior optimized results; for example, reducing both the shaft weight and transmitted forces of the bearing for rotor-bearing systems under critical speed constraints.