Developing an Hourly Multi-reservoir Optimization System Adaptable to Extreme Weather Conditions Using Multi-objective Optimization, Fuzzy Logic, and Fuzzy TOPSIS

Abstract Hydropower offers notable advantages due to its clean energy source and quick response capabilities, making it essential for stabilizing the power grid. However, it also serves multiple purposes, such as municipal water supply and agricultural irrigation. Recently, extreme weather patterns, especially heavy rainfall in island climates, have made reservoir power generation planning more complex. Short-term intense rainfall often causes reservoir inflows to exceed power generation needs, leading to flood discharge decisions. Additionally, smaller reservoirs are highly susceptible to rapid water level fluctuations, and traditional daily or longer planning methods severely limit the search space for optimization. To tackle these issues, this paper presents a two-stage hourly multi-reservoir optimization system. This system facilitates hourly power generation planning across the entire basin while ensuring that overall power generation targets are met. The proposed system includes two stages: the first uses ev-MOGA and cost functions to produce a Pareto set of solutions based on flood risk factors and water use. The second-stage employs fuzzy TOPSIS, along with expert-weighted adjustments, to select the best solution. The proposed system also integrates inflow predictions to improve accuracy and address uncertainties, with its effectiveness demonstrated in Taiwan’s Dajia river basin.