An improved two-step parameter adjustment method for the optimization of a reservoir operation function model based on repeated principal component analysis and a genetic algorithm

This paper presents an improved two-step parameter adjustment method for the construction of a reservoir operation function model, by using repeated principal component analysis (PCA) and a genetic algorithm (GA) to optimize parameters in conventional multiple regression models. The first step is to use repeated PCA, to exclude the co-linear parameters in a multiple regression expression reflecting relationships among possible impact factors of reservoir operation, so as to form an initial reservoir operation function model. The second step is to use a GA to optimize the model constructed in the first step, and to compare its effects with other regression methods. The results show that the proposed reservoir operation function model can produce better results, which correlate water volume for power generation, input discharge, water level, and ecological flow. Compared with established scheduling schemes, the optimized scheme increases the water volume for power generation by 1.06 × 109 m3/yr, and the optimized result generates an increase in economic benefits of 3.22 × 107 yuan/yr (i.e., 4.69 × 106 USD/yr).