An Improved Efficient Ordering Evolutionary Algorithm For Many-objective Optimization And Its Application

Evolutionary algorithms are a kind of random search algorithms that simulate natural selection and natural evolution.It is widely used for their strong ability of global optimization.In recent year,it has been a hot research in the field of evolution,while using evolutionary algorithms to solve multi-objective optimization problems.Evolutionary algorithms have advantages to solve optimization problems with few objectives.However,in real life,optimization problems often involved many objectives(at least four).In this case,many-objective evolutionary algorithm came into being.Compared with the multi-objective evolutionary algorithm,the most important difficulty of many-objective evolutionary algorithms is that the number of non-dominated solutions increases rapidly with the increase of the number of the objective.Normally,most exiting multi-objective evolutionary algorithms using Pareto-based sorting method to choose better solutions,this leads to the decrease of selective pressure and convergence speed.In order to relieve the difficulty,a new algorithm is proposed called an improved efficient ordering evolutionary algorithm for many-objective optimization by making a research on lots of many-objective algorithms and sorting methods.This algorithm is applied to solve nurse rostering problem by adjusting the crossover and mutation operator.The main work of this paper is as follow:1.The current situations of evolutionary algorithms and many-objective evolutionary algorithms are introduced systematically,and the difficulties of the many-objective evolutionary algorithms are introduced in detail.Then,how to solve the difficulties of many-objective evolutionary algorithms with the methods proposed by the researcher in the filed of evolution is introduced.2.An improved effective ordering evolutionary algorithm for many-objective optimization is proposed.Firstly,an improved effective ordering algorithm to reduce the number of the non-dominated solutions is proposed,and it increases the convergence speed of the algorithm.Secondly,a new calculating crowding degree method with hypervolume is proposed,which is mainly used to calculate the crowding degree of a point in the objective space.Thirdly,the hypersphere polar coordinate search algorithm is used,which produces some points evenly in the boundary of the objective space and the sparse area,in order to increase the diversity of the population.Finally,a new crossover operator is proposed toincrease the diversity of the population.Through the test of DTLZ1,DTLZ2,DTLZ3 and DTLZ4 having four to eight objectives,the experiment results show that the algorithm is proposed in this paper is superior to NSGA-II algorithm and Preference Order Genetic Algorithm(POGA)algorithm.on the performance of convergence.3.An improved multi-objective evolutionary algorithm to solve nurse rostering problem is proposed.The algorithm regards the more important constraint as an objective,and the other constraints are regarded as another objective using the average weighted method.And then establish the mathematical model of multi-objective optimization problem.Finally,the algorithm is used to deal with the nurse rostering problem,which provides a new idea for nurse rostering problem.The algorithm is different from other simple optimization algorithms.Because multi-objective evolutionary algorithm can get the optimization solution set every time,the decision maker can select the appropriate solution according to his own need.