Towards Many-Objective Optimization: Objective Analysis, Multi-Objective Optimization and Decision-Making

Abstract

This paper presents a tri-level many-objective optimization (TLMaO) approach to provide a final solution for many-objective optimization problems (MaOPs). In this approach, the proposed objectives' number reduction (ONR) method is utilized as the first level to select the most conflicting objectives for the second level to optimize. The second level outputs a set of Pareto-optimal solutions using the multi-objective optimization algorithm, however, a unique solution must be selected for real world problems. Therefore, we propose an improved entropy weight (IEW) method for decision making as the third level to determine the final solution. The effectiveness of the ONR and IEW method is first demonstrated on test problems. Then, the features and efficacy of the proposed TLMaO approach are investigated on a real world problem, the many-objective optimization of power flow (MaOPF). The simulation results verify that when compared with a general method used for MaOPs, our TLMaO approach can offer more competitive and robust solutions.

Keywords

• Objectives’ number reduction
• Spearman’s rank correlation coefficient
• many-objective optimization
• decision making
• entropy weight