文摘
In everyday life, there are plentiful cases that we need to find good solutions such that risk, cost and many other factors are to be optimized. These problems are typical examples of multi-objective optimization problems. Evolutionary algorithms are often employed for solving it. Due to the characteristics of learning and adaptability, self-organization and memory capabilities, one of the biological inspired AI methods 8211; artificial immune systems (AIS) is considered to be a class of evolutionary techniques that can be deployed for solving this problem. This paper aims to propose a new AIS-based framework focusing on distributed and self-organization characteristics. Population of solutions is decomposed into sub-populations forming clusters. Sub-populations in each cluster undergo independent evolution processes. These clusters are then combined and re-decomposed. The proposed mechanism aims to reduce the complexity in the evolution processes, enhance the exploitation ability and achieve quick convergence. It is evaluated and compared with representative algorithms.