A Developed NSGA-II Algorithm for Multi-objective Chiller Loading Optimization Problems

During recent years, for its simplicity and efficiency, the non-dominated-sorting algorithm (NSGA-II) has been widely applied to solve multi-objective optimization problems. However, in the canonical NSGA-II, the resulted population may have multiple individuals with the same fitness values, and which makes the resulted population lack of diversity. To solve this kind of problem, in this study, we propose a developed NSGA-II algorithm (hereafter called NSGA-II-D). In NSGA-II-D, a novel duplicate individuals cleaning procedure is embedded to delete the individuals the same fitness values with other ones. Then, the proposed algorithm is tested on the well-known ZDT1 instance to verify the efficiency and performance. Finally, to solve the realisitc optimization problem in intelligent building system, we select a well-known optimal chiller loading (OCL) problem to test the ability to maintain population diversity. Experimental results on the benchmarks show the efficiency and effectiveness of the proposed algorithm.