Hydropower Scheduling of Cascaded Reservoirs Using a Hybrid Gurobi–PSO Algorithm

Abstract

Optimal operation of short-term power generation solutions for cascade hydropower stations is an important part of current reservoir operation and management. Existing population intelligence optimization algorithms exhibit limitations including susceptibility to the initial population and early premature convergence when applied to this field. To boost efficiency and accuracy in meeting the generation needs of cascade hydropower stations, this paper proposes a Gurobi–PSO algorithm to address the characteristics of the particle swarm optimization (PSO) algorithm by integrating the fast construction capability of the Gurobi solver, the hybrid initialization strategy and the parallel computation framework, and applies it to the short-term power generation optimization operation for cascade hydropower stations on the Lower Jinsha River. Results show that, compared to the standard PSO, the Gurobi–PSO algorithm produces an initial population with superior distribution and convergence characteristics, reducing the average solving time by 9.6 seconds and improving computational efficiency by approximately 57%. Furthermore, it increases the total average power generation of the cascade hydropower stations by 10,596.08 MWh, corresponding to a 1% improvement in generation. When compared to the dynamic programming successive approximation (DPSA) algorithm, the proposed method reduces computation time by 235.09 seconds and increases efficiency by approximately 97%, provided that the computed power generation is not much different from the DPSA algorithm. The Gurobi–PSO algorithm provides a more efficient, stable, and practical optimization tool for the short-term dispatch of cascade hydropower stations.