Research And Application Of Discrete Multi-Objective Optimization Based On Genetic Algorithms

There are many multi-objective optimization problems in the field of scientific research and practical engineering.Multi-objective optimization technology can compromise the conflicting objectives to achieve the optimal state for specific requirement.It makes a great influence on some complex scientific research and practical engineering problems.So,it is importance to develop multi-objective optimization algorithms.This thesis takes discrete multi-objective optimization problems(DMOP)as the object of the research.The following results have been achieved:A genetic algorithm with better performance is obtained.Aiming at the shortcomings of the NSGA2 in convergence,distribution and computational efficiency,a new genetic algorithm is improved.In terms of convergence,a local search strategy based on density is introduced.In terms of distribution,the cyclic congestion ranking strategy is introduced.In terms of computational efficiency,the Pareto frontier hierarchical strategy on demand is introduced.The convergence,distribution and efficiency of the algorithm are improved.A multi-objective discrete genetic algorithm that can process decision variables of discrete point sets is proposed.A related study is made on the discrete multi-objective optimization problem in which the decision variables are in the form of discrete point sets.Aiming at this kind of discrete multi-objective optimization problem,the discrete variable processing method based on minimum Euclidean distance value strategy is proposed to make the algorithm can be optimized in the discrete solution space.To some extent,it overcomes some shortcomings of the traditional algorithm.It effectively improves the optimization efficiency of the algorithm.A multi-objective discrete genetic algorithm that can process decision variables of discrete interval sets is proposed.A related study is made on the discrete multi-objective optimization problem in which the decision variables are in the form of discrete interval sets with equality constrain.Aiming at this kind of discrete multi-objective optimization problem,a discrete variable processing method based on in infeasible solutions for improved two-layer solution repair strategy is proposed to make the algorithm can deal with the discrete interval optimization problems with the equality constraint quickly and effectively.To some extent,it overcomes some shortcomings of traditional algorithms.It effectively improves the optimization efficiency of the algorithm.The above discrete genetic algorithms are applied to two practical engineering problems and better optimization results are obtained.