A Parallel Multi-objective Optimization Algorithm Based on Coarse-to-Fine Decomposition for Real-time Large-scale Reservoir Flood Control Operation

Abstract

Reservoir flood control operation (RFCO) is a multi-objective optimization problem with a long sequence of correlated decision variables. It brings big challenges to large-scale multi-objective optimizers which were generally developed based on the divide-and-conquer strategy. For solving large-scale RFCO problem, a novel coarse-to-fine decomposition method is developed and combined with the algorithmic framework of multi-objective evolutionary algorithm based on decomposition (MOEA/D), giving rise to the proposed pCFD-MOEA/D algorithm. The pCFD-MOEA/D algorithm first divides the original RFCO problem into a sequence of sub-problems from coarse to fine scale with different scheduling time intervals. Then all sub-problems are optimized simultaneously and communicate at set intervals. Experimental results on three typical floods at Ankang reservoir have demonstrated that the proposed pCFD-MOEA/D can successfully obtain the elaborate hourly schedule schemes in real time and outperforms the compared algorithms.