The Study Of The Archive In The Multi-objective Optimization Algorithms

Multi-objective optimization algorithm uses archive to store all optimal solutions in the process of running. The introduction of the archive greatly improves the performance of multi-objective optimization algorithm. Traditionally, the research of archive is focus on the issue of how to maintain the diversity, which uses various density estimation techniques to delete the redundant solution. However, the archive stores all of the optimal information, how to use of this information effectively to enhance the ability of algorithm to overcome the premature convergence is the main research content in this paper.At first, this paper studies the change process of archive size, based on this, a simple method which be used to determine the status of the algorithm running is proposed, and then the parameters can be adaptive adjusted in the running. The experimental results shows that the parameter adaptive strategy based on this determination can effectively improve algorithm performance. In addition, this article has mainly studied the update operation of archive, in order to effectively use the information stored in the archive, this paper puts forward using differential evolution and mutation operation to update the archive. By comparing with several famous algorithms, the experiment result shows that the update operation can obviously improve the ability of the algorithm to jump out of local optimal.