Performance of genetic algorithms with different selection operators for solving short-term optimized reservoir scheduling problem

The tournament operator genetic algorithm (TGA) often shows poor convergence and easily gets trapped in a local optimum when solving optimized reservoir scheduling problems. The selection operation is the most important operation determining an algorithm’s convergence; therefore, this study proposes a proportional reproduction selection-based operator genetic algorithm (RGA) and a steady-state reproduction selection-based operator genetic algorithm (SGA) as alternatives to TGA. This study used TGA, RGA, and SGA to solve the maximum power generation model for the Gezhouba hydropower station, the largest runoff hydropower station in the world. Then, by using the maximum hydropower station output under a given typical runoff scenario as the optimization criterion, this study evaluated the optimized solution performance of the GA using different selection operators. The results show that TGA, SGA, and RGA can be applied to solve the short-term reservoir scheduling model. As the number of iterations increases, the hydropower station output optimized by these three GAs increases. Based on the maximum power generated by the Gezhouba hydropower station, SGA and RGA provide better results than TGA, and between SGA and RGA, the former provides better results.