Multi-objective Brain Storm Optimization Based on Estimating in Knee Region and Clustering in Objective-Space

摘要

The knee region of the Pareto-optimal front is important to decision makers in practical contexts. In this paper, a new multi-objective swarm intelligent optimization algorithm, Multi-objective Brain Storm Optimization based on Estimating in Knee Region and Clustering in Objective-Space (MOBSO-EKCO) algorithm is proposed to get the knee point of Pareto-optimal front. Firstly, the clustering strategy acts directly in the objective space instead of in the solution space, which suggests the potential Pareto-dominance areas in the next iteration more quickly. Secondly, the estimating strategy is used to discover the knee regions, which are the most potential part of the Pareto front. Thirdly, Differential Evolution (DE) mutation is used to improve the performance of MBSO. Experimental results show that MOBSO-EKCO is a very promising algorithm for solving these tested multi-objective problems.