A cellular automata approach for the hydro-power operation of multi-reservoir systems

Abstract

A cellular automata (CA) method is proposed in this paper for the optimal operation of hydropower multi-reservoir systems. The beginning and the end of the operation periods are taken here as the CA cells leading to the storage volumes of the system being defined as the cell state. This choice naturally leads to a cell neighbourhood defined by the previous and next periods of the underlying cell. The objective function and constraints of the underlying optimal operational problem are projected on each cell to arrive at the local updating rule of the CA method. The resulting updating rule is, therefore, defined by an optimisation sub-problem of a size equal to the number of reservoirs in the system which is subsequently solved by NLP technique to get the updated values of the cell states, storage volumes of the reservoirs in the system. The efficiency and effectiveness of the proposed CA method is tested against two multi-reservoir systems, namely four- and ten-reservoir problems over 12, 60 and 240 monthly operation periods, and the results are presented and compared with those obtained by two of the most commonly used continuous heuristic search methods, namely genetic algorithm (GA) and particle swarm optimisation (PSO) algorithms. The results show that the proposed CA method is more efficient and effective than the GA and PSO algorithms, in particular for the solution of large-scale multi-reservoir hydropower operation problems.