Research On Many-Objective Evolutionary Algorithm Based On Reference Point

In recent years,science and technology have developed rapidly and various optimization problems have emerged in many fields such as engineering practice,production and life.Evolutionary algorithm,as a representative intelligent optimization algorithm,has attracted the attention and research of many scholars because it can solve the performance of discontinuous and non-conductive functions and other superior performance.With the further application of engineering practice and scientific research,many-objective optimization problems have gradually entered the field of view of scientific researchers.In the many-objective optimization problem,as the size of the objective function increases,the non-linearity increases,and the overall complexity also increases.There is also an increasing demand for intelligent algorithm's solution performance.In this paper,we study the many-objective optimization problems,introduce the relevant definitions of many-objective evolutionary algorithms,describe the difficulties encountered by current multi-objective evolutionary algorithms when dealing with many-objective problems,and discuss the high dimensions that have been proposed in recent years.The multi-objective evolutionary algorithm is classified and introduced in detail.In order to better solve the many-objective optimization problem,the classical non-dominated sorting genetic algorithm and the many-objective evolutionary algorithm based on the reference point were deeply studied.Through the preliminary improvement of the reference point generation strategy,it is combined with a fast non-dominating genetic algorithm to form a new genetic algorithm that can solve many-objective problems.Through the improvement of reference point generation strategy,the improved reference point generation strategy is combined with the many-objective evolutionaryalgorithm based on reference points.The test results of the standard test function show that the improved reference point evolution algorithm can effectively solve the many-objective problem.