Research And Application Of New Non-Dominated Individual Sorting In Multi-objective Evolutionary Algorithms

In the field of scientific research and engineering applications, many problems can be abstracted as multi-objective optimization problems. Due to the mutual restriction between the optimization targets in multi-objective problems, a target performance improvement may cause other target performance degradation,what need to deal with more than one optimization problem is known as multi-objective optimization problem. Multi-objective evolutionary algorithm for its intelligence, versatility, powerful global search capability to become an important algorithm for solving multi-objective optimization problem.Based on the important evolutionary dynamics of Pareto strategy, NSGA-II, SPEA2 and other algorithms plays a decisive role in the evolution of population, which will affect the final solution set. Therefore, the study of individual sorting strategy of multi-objective evolutionary algorithm has become a hot research field in evolutionary computation. Based on the above facts, a new type of non-dominated sorting strategy is designed for the problems existing in the individual sorting strategy in multi-objective evolutionary computation, and the strategy is applied to the study of NSGA-II, SPEA2 and PESA-II algorithms.Through the mainstream domestic and foreign literature research and analysis, this paper proposes a new non-dominated individuals sorting policy, Simultaneously,an improved NSGA-II algorithm based on the new non-dominated sorting individual policy is designed. Last,For useful exploration of new non-dominated individuals sorting policy,we proposed new policy to SPEA2 and PESA-II algorithm.1. Firstly, we verified from two different perspectives on the theoretical analysis and experimental NSGA-II algorithm individual crowding distance sorting strategies are analyzed, it demonstrates the inadequacies of the policy making individual choices sort exists. Then we put forward a new strategy based on individual sorting clustering algorithm. Then, an adaptive hybrid non-dominated sorting strategy based on individual characteristics of NSGA-II algorithm, while the embedded policy to the NSGA-II algorithm proposed a model based on adaptive hybrid non-dominated individuals sorting policy improved NSGA-II algorithm. The experiments on six multi-objective benchmark functions show that the NSGA-IIh algorithm can acquire 83% of optimal IGD value, and the corresponding two-tailed t-test results at a 0.05 level of significance are remarkable.2. In order to further verify the new individual sorting strategy that proposed is scalability and the distribution and convergence of the solution set of the new algorithm, this paper relies on SPEA2, PESA-II algorithm model, the new research policy to the individual sorting the two algorithms are proposed based on the new improved algorithm individual sorting policy and Finally the new algorithm is proved can improve the distribution and convergence of the solution set. At the same time, it proves that the new non-dominated individual sorting policy has strong expansibility.