关键词: ＮＳＧＡⅡ, 拥挤度距离, 收敛性, 水库流量控制系统

Abstract: NSGA-Ⅱ is a classical multi-objective optimizer. However, the crowding distance in NSGA-Ⅱ has the drawback of ineffectively identifying promising individuals. Regarding this issue, this paper firstly analyzes the essential reason for this drawback, and further proposes a new evaluation strategy which is based on the sum of objectives. Based on the proposed evaluation strategy, this paper then proposes an improved NSGA-Ⅱ. To assess the improved NSGA-Ⅱ, this paper uses DTLZ test suite and multiple state-of-the-art optimizers to analyze the improved NSGA-Ⅱ. The experimental results indicate that the improved NSGA-Ⅱ outperforms most state-of-the-art methods. Besides, this paper further applies the improved NSGA-Ⅱ to a water reservoir flow control system problem. The experimental results demonstrate the outstanding performance of the improved NSGA-Ⅱ on practical engineering problems, indicating the practicality of the improved NSGA-Ⅱ.