Medium-Term Multiobjective Operation Mode of Cascade Reservoirs Using Multisource Information

Abstract

The incorporation of information on the current system states and forecasts makes reservoir operations more efficient; however, forecast uncertainties often occur. Therefore, excellent reservoir operations should integrate forecast uncertainties and current system states to achieve trade-offs among the objectives. This study proposed a medium-term multiobjective operation (MTMOO) mode, which combines the coupled rainfall forecast (CRF) and the current system state as multisource information to intuitively guide the multiobjective scheduling decision-making of cascade reservoirs. The CRF in different forecast periods was determined to mitigate forecast uncertainties. The MTMOO mode was formulated by establishing a multiobjective optimization model and then using the decision tree algorithm. For verification, the CRF, 5-day rainfall forecast, and observed rainfall were simulated using the MTMOO mode, and the results were compared to those of conventional scheduling. The case study shows that the electricity generation and water supply benefits under the MTMOO mode using CRF are much greater than those in conventional scheduling and slightly larger than in the MTMOO mode that directly uses rainfall forecast. This confirms the applicability and superiority of implementing the MTMOO mode with CRF as an input in multipurpose reservoirs.