Daily optimized model for long-term operation of the Three Gorges-Gezhouba Cascade Power Stations

Abstract

This paper presents the step-by-step genetic algorithm based on artificial intelligence guidance and builds a long-term daily optimized operating model for the Three Gorges-Gezhouba Hydropower Complex with single generating set as the based operating unit. Actual operating data from 2004 to 2006 are used to verify the model, and results show that the simulation accuracy determined by measuring the total amount of cascade power generation reaches 99.66%. Statistic hydrological data of normal years and actual data of three years of 2004–2006 are respectively used to perform an optimized prediction of the power generation process and benefits in future when the water stored in the TGP Reservoir reaches 175 m level and power generation benefits under different operation modes, such as delayed subsiding water level, advance water storage, and adopting of different flood-limited water levels, are forecasted. In the case of years with normal inflows, the total amount of cascade power generation running on current specifications reaches 107500 GWh per year. If the commencement of water storage after the flood season is moved forward by 20 days, the amount of power generation can be increased by 3400 GWh per year. If the limited water level in the flood season is raised by three to five meters, the amount of power generation can be increased by 1600 to 3200 GWh per year. If the commencement of water storage is moved forward while the maximum water level allowed in the flood season is raised, the amount of power generation can be increased by 6400 GWh per year.