Research On Multi-objective Optimization Algorithm Based On Harmony Searching

At present,the multi-objective optimization method is mainly based on Pareto mechanism,while its core is to obtain the Pareto frontier,if not,then to get the non-inferior solution which is closed to the Pareto frontier as much as possible,this is the compromise process of objective.Next,it's necessary to extract the final decision information according to the requirements of problems to be solved or the preference of decision makers,then determine a solution as the final "optimal solution" from the non-inferior solution,this is the decision-making process.In this research,we focus on the core work of optimization,to introduce and study a new intelligent evolutionary algorithm-harmony search algorithm,in order to achieve the object solving.The prominent advantages of the harmony search algorithm are simple,easy to achieve,and to mix with other algorithms and mechanisms.But the harmony search algorithm also exists the weaknesses such as "precocity"convergence,having low convergence speed in the late evolution and so on,especially it will be very hard to search the global optimal solution to the complexed optimization problems.According to the above,in this research,firstly the harmony search algorithm is improved,which optimizes the initial solution through chaos mechanism,improves the quality of solution set,and makes adaptive adjustment to the controls parameters,in order to make it more suitable for searching the different requirements for parameters in all stages,which improves the performance of algorithm.On this basis,combine the harmony search algorithm with non-dominated sorting thought general in multi-objective theoretical framework,then propose the multi-objective harmony search algorithm based on chaotic adaptive(MOCAHS).Simulation study,select 7 standard test functions to evaluate the algorithm performance,which verified the excellent properties of MOCAHS algorithm.By analyzing and verifying the performance of MOCAHS algorithm,we may find that as a kind of single strategy group of intelligent algorithm,it still exists the weakness,which could not focus on and balance local exploration ability and global development ability of the algorithm in the same phase and at the same time,the more objectives,the weakness is more obvious.In order to solve the weakness,in this research we put forwardthe multi-objective harmony search algorithm(MOMPHS)based on co-evolution of multi-population,the algorithm introduces the co-evolutionary mechanism,designs three populations with different evolutionary strategy,makes it have the abilities of global development,local exploration and balance,the three populations exchange information,cooperate with each other,and co-evolution to find the Pareto frontier.At last,through simulation experiment,compared the MOCAHS and MOMPHS algorithm with other classical evolutionary algorithm,verified the algorithm of this research played very well on dealing with the multi-objective optimization problem,which achieved the desired effect.