Research On Super Multi-objective Optimization Algorithm Based On Swarm Intelligence

In the daily life, there exist a lot of complex optimization problems. What's more, these optimization problems always consist of more than one optimization objectives, which contradict and restrict with each other. As it is difficult to achieve an optimum for all the objectives simultaneously, the conventional operational research methods can not solve them well. By simulating the biological behaviors, evolutionary algorithms inspired from the biological theory have the advantages in solving these optimization problems. Especially, decomposition-based evolutionary algorithm decomposes multiobjective optimization problems into a set of single-objective optimization problems using the mathematical programming method, which provides a novel and efficient approach to solve the complex problems. Therefore, it attracts more and more attentions from different application fields of researcher. Moreover, many-objective optimization problems also have received a wide attention. Due to its simple concept, easy implementation, and fast convergence speed, multi-objective particle swarm optimization can avoid the dimension disaster problems in solving many-objective optimization problems.This paper studies evolutionary algorithms mainly from the view of optimization problems. In this paper, we focus the main work on decomposition-based multiobjective evolutionary algorithm for solving multi-objective optimization problems and multi-objective particle swarm optimization algorithm for solving many-objective optimization problems. We propose a decomposition-based multiobjective evolutionary algorithm with adaptive operator selection(AMOEA/D) and a novel particle swarm optimization algorithm for many-objective optimization problems(NMPSO). This paper firstly introduces the background of multi-objective algorithms based on swarm intelligence research and the optimization problems. Then, this paper respectively analyzes the present situation of multi-objective evolutionary algorithm based on decomposition and particle swarm optimization algorithms. Moreover, we also put forward two new algorithms for the above two kinds of optimization problems. The main work of this paper is described as follows:1. When solving multi-objective optimization problems, this paper proposes an adaptive decomposition-based algorithm(AMOEA/D), which consists of an adaptive operator selection for speeding up the convergence and enhancing the search ability. Moreover, the stable matching method is also adopted for the stable matching states where the subproblems and solutions can match with each other. According to the stable matching method, the algorithm can keep the diversity of the population.2. When solving many-objective optimization problems, this paper proposes a new multi-objective particle swarm optimization algorithm, which consists of a new crowding distance assignment for avoiding the dimension disaster problems. The main principle of this method is to preserve the better particles and to delete the worse particles, while the trade-off particles must have a punishment on the population diversity. This new multi-objective particle swarm optimization algorithm can be simply implemented, whose effectiveness is also significant. Meanwhile, the evolutionary operator is used to disturb the archive individuals, which can keep the diversity of the solution. This new strategy of updating the speed of the particles makes the convergence speed faster.