Research on the optimal operation calculation model of large-scale hydropower reservoir systems using a hybrid algorithm

Abstract

Researchers have applied many optimization techniques to different reservoir systems, the most popular of which are dynamic programming (DP) techniques and genetic algorithms (GA). Still, these techniques are not effective for multi-reservoir systems. The discrete difference dynamic programming algorithm (DDDP), a variation of dynamic programming, is the solution of choice for reservoir system problems. Still, its drawback is that it must find the optimal initial trajectory. If recursive equations are used to find the optimal trajectory, this solution also encounters the problem of dimensionality of the dynamic programming problem. Meanwhile, GA can find the optimal solution from random solution sets. Still, its disadvantage is that finding the global optimal solution is tough, while calculation time is also a limitation of GA for significant system problems. Combining GA and DDDP for multi-reservoir problems is an effective method for these two algorithms. The case study is the system of five cascade hydropower reservoirs in the Da river basin in Northern Vietnam. Keywords: Reservoir system, genetic algorithm, discrete differential dynamic programming, cascade hydropower