Multi-objective Reservoir Optimal Operation Based on GCN and NSGA-II Algorithm

摘要

The problem of reservoir flood control operation is a multi-objective optimization problem. In order to solve this problem, a new algorithm based on graph convolution neural network and fast non-dominated sorting genetic algorithm II (GCN_NSGA-II) is proposed in this paper. When NSGA-II algorithm is used to solve multi-objective optimization problems, its convergence speed will slow down when the iteration reaches a certain algebra. In order to speed up the convergence speed, we use genetic algorithm to simulate the reproduction process of biological population. There is a relationship between parents and offspring. By means of group coding, the tree structure of the parents and children is transformed into a graph structure, and the GCN is trained and the graph nodes are classified, and the Pareto solution set can be obtained more quickly. In order to further ensure the integrity and uniformity, the NSGA-IIalgorithm is used to adjust it. The performance index IGD is used to measure the algorithm in the process. This algorithm speeds up the convergence speed and ensures the uniformity of Pareto solution set. The effectiveness of the algorithm in this paper is verified on the Xiaolangdi Reservoir flood control and dispatching problem. Compared with the NSGA-II and NSGA-DE algorithms, for the same index value, the number of iterations of the GCN_NSGA-II algorithm is significantly reduced and the convergence speed is significantly accelerated.