Research And Aplication Of The Hybrid Genetic Algorithm To Solve The Multi-objective Optimization Problem

In recent years, it has become evident that the concentration on a sole genetic algorithm is rather restrictive. A skilled combination of concepts of the multi-objective genetic algorithm and other optimization techniques, also called hybrid hybrid multi-objective genetic algorithms, can provide a more efficient behavior and a higher flexibility when dealing with real-world problems. In this paper, on the basis of analyzing of the NSGA2, SPEA2, MOEA/D and PLS algorithms, we implemented some corresponding improvements for the deficiencies of these algorithms. And we designed two PLS-based local optimization strategys, which are combined to the NSGA2, SPEA2and MOEA/D algorithm with effect, respectively, proposed the impoved NSGA2algorithm-INSGA2, the impoved SPEA2algorithm-ISPEA2and the impoved MOEA/D algorithm-IMOEA/D. According to the characteristics of optimization mechanism of the SPEA2and MOEA/D algorithm, and combined to the new evolutionary strategies by the co-evolution way, we proposed other hybrid multi-objective genetic algorithm-HDMOGA. Then, we used the proposed hybrid muti-objetive genetic algorthims to optimize the different scale and represent-ativeness tests of the multicast QoS-based guaranteed rounting and wavelength assignment problem (MQRWA) and the service selection based on QoS constraint problem (QSS), which are the NP-hard muti-objective optimization problem. In this paper, we proposed respectively the mathematical optimization mode, and maded a specific crossover operator, mution operator and PLS optimization strategies for the two optimization problems. Specially, for MQRWA problem, we proposed two repair strategies of light tree (RDFS and PRRA) and initialization strategie based on Prim minimal spanning tree. For QSS problem, we proposed the preference based initialization strategies. Finaly, we use different indicators to analyze the algorthm performances from various perspectives. The experiments show that the improved hybrid muti-objetive genetic algorithms demonstrated their superiority in terms of the global optimization and local optimization capabilities.The field of hybrid muti-objetive genetic algorithm is a research hot, but many approaches are pre-mature and a substantial amount of further research is necessary in order to develop clearly structured hybrid muti-objetive genetic algorithm. So, this paper’s work will not only enrich and push forward the studies of the related areas in both theoretical and technological aspects, but also proposed a new means to optimize the MQRWA and QSS problems.