The Research And Improvement On Selection Method Of Multiobjective Optimization Genetic Algorithms

Multi-objective optimization is a difficult problem and a research focus in the fields of science and engineering. Classical multi-objective optimization methods have several shortcomings in solving high dimension, multi-model problems. In order to solve these problems, researchers have developed many multi-objective optimization genetic algorithms based on Simple Genetic Algorithm (GA). And, as the capability of searching the global best solutions rapidly, Genetic Algorithm has become a research area with increasing importance.The main works of this paper are as follows:1) The problems of multi-objective optimization and the current state of the research on these problems are introduced. Meanwhile, the basic theories and concepts of genetic algorithm are also systematically presented.2) In addition, this paper also focuses on the problem of how to rank solutions into order by fitness within genetic algorithms for multi-objective optimization problems. Six methods for solving this multi-fitness ranking problem are described in details and applied to three test functions for comparison. Results show that all methods allow the generated of Pareto-optimal solutions, but all have different distributions of solutions within the Pareto-optimal ranges. The bias of each distributions and the resulting quality of solutions generated by each method is examined and compared. This paper concludes the reason for this phenomenon.3) Furthermore, a new ranking method of multi-objective optimization genetic algorithms based on the nonlinear selective strategy is presented to solve the problems exist in the usual ranking methods which have been widely used. This new strategy improved the performance of the usual ranking methods, which could get better distribution and find better non-dominated fronts that converge near to the global Pareto-optimal front in few times.