Author(s): Shuang Zhang; Tifeng Zou; Shi Feng

Linked Author(s):

Keywords: Digital twin; Small hydropower; Machine learning; Intelligent operation; Ecological flow

Abstract: Small hydropower plants (SHPs) are essential for renewable energy generation and regional water management. Conventional monitoring and operational approaches often suffer from limited real-time data integration, insufficient predictive capability, and delayed response to disturbances, leading to reduced energy efficiency and increased mechanical fatigue. To address these issues, this study presents a digital twin framework for SHPs, integrating real-time sensing, hybrid modeling, and intelligent control. The system combines physics-based and data-driven models, with sensor networks collecting hydrological, meteorological, and equipment status data processed via a cloud-edge architecture. Machine learning algorithms predict streamflow, turbine performance, and potential faults, while a model fusion strategy reconciles physical and data-driven predictions for accurate state estimation and operational decision support. The framework was deployed at a pilot SHP, achieving 5–7% mean absolute error in streamflow and power prediction over six months. Real-time monitoring and optimized control increased overall energy efficiency by 3–5% and maintained continuous ecological flow compliance. The proposed digital twin approach demonstrates strong potential for scalable and intelligent management of SHPs enhancing predictive capability, operational efficiency, and environmental sustainability. ,

DOI: https://doi.org/10.64697/iahr.proc.hic2026.47

Year: 2026