A hybrid NSGA II-CNN-LSTM model for optimal operation of hydropower reservoirs based on the day-ahead electricity price forecast

Hydropower is one of the renewable and affordable energy sources, though it may cause environmental impacts that require sustainable management. To ensure sustainability, hydropower producers must adopt a multi-dimensional approach to reservoir operation, balancing reservoir conditions with the three pillars of sustainability: environmental requirements, societal energy demand, and electricity market prices. In hydropower plants (HPPs), both the volume of water released and the timing of release significantly impact profitability. Consequently, HPP operators must strategically forecast electricity prices while planning water discharges and power generation to align with peak demand periods. This study employs a novel hybrid model combining the CNN-LSTM for forecast of daily peak electricity prices and NSGA-II optimization tool for optimal operation of the Karun-IV hydropower reservoir. The hybrid model objective is maximizing the total revenue of the system while the minimum daily electricity generation over a 10-year operational period is maximized. The results demonstrate successful energy arbitrage where the reservoir acts as a virtual battery by storing water during low electricity demand periods and releasing it when demand for energy is higher. Results indicate that the combined NSGA II-CNN-LSTM tool can enhance the HPP income about 10.5% compared to the historical recorded data. This highlights the tangible benefits of integrating accurate market forecasts with the optimization tools, offering a valuable tool for energy operators.