Multi-objective optimization problem has a wide range of applications in the real world. It often involves incommensurable and competing objectives. So there was not single solution vector that made all the objectives optimal. As a result, it was difficult for optimizing the multi-objective problem.Evolutionary algorithms were inspired from the ideas from natural evolutionary processes. Due to its intrinsic parallelism, self-organizing, adaptation and self-learning intelligent properties, evolutionary algorithms have large potential to solve multiple objectives optimal solutions. The multiple objectives optimization and decision-making has become an important research area of evolutionary algorithms in recent years.Firstly, the research status and methods of evolutionary multi-objective optimization were systematically elaborated and summarized. Then, some state of art multi-objective evolutionary algorithms were analyzed for their advantages and drawbacks, based on which the non-dominated sorting differential evolution algorithm was proposed for the multi-objective optimization. The proposed algorithm inherited the fast non-dominated sorting approach and the diversity preservation method of NSGA-â…¡. At the same time, it made use of the simple but efficient differential way for obtaining the offspring in stead of the simulated binary crossover and polynomial mutation. For the further improvement of the performance of the proposed algorithm, the direct convergence and spread information was incorporated in the differential method. The convergence information help to the improvement of the convergence speed of the algorithm. The spread information was used to optimize the distribution of the solutions from the algorithm.Matlab was used for the numerical simulation. The experimental results demonstrated the solution obtained by the non-dominated sorting differential evolution algorithm had better convergence and distribution properties than NSGA-â…¡. |