Feng et al. (2026) Artificial intelligence-based response surface stochastic dynamic programming for long-term operation rule derivation of cascade hydropower reservoirs

Identification

• Journal: Journal of Hydrology
• Year: 2026
• Date: 2026-08-10
• Authors: Zhong-kai Feng, Xuan Chao, Yang Xiao, Hong‐Wu Tang, Wenjing Niu, Tao-tao Zhang, Ling-zhong Kong
• DOI: 10.1016/j.jhydrol.2026.136216

Research Groups

• State Key Laboratory of Water Disaster Prevention, Hohai University, Nanjing, China
• College of Hydrology and Water Resources, Hohai University, Nanjing, China
• School of Environmental Science and Engineering, Suzhou University of Science and Technology, Suzhou, China
• Key Laboratory of Hydrologic-Cycle and Hydrodynamic-System of Ministry of Water Resources, Hohai University, Nanjing, China
• School of Civil Engineering and Transportation, South China University of Technology, Guangzhou, China
• Bureau of Hydrology, ChangJiang Water Resources Commission, Wuhan, China

Short Summary

This study presents a response surface-based stochastic dynamic programming (RSSDP) algorithm for long-term optimal power generation operation of hydropower reservoirs. The RSSDP method significantly improves computational efficiency while maintaining scheduling accuracy comparable to traditional stochastic dynamic programming.

Objective

• Develop an efficient and feasible technical pathway for the formulation of long-term optimal operation rules for cascade reservoirs using response surface-based stochastic dynamic programming (RSSDP).

Study Configuration

• Spatial Scale: Cascade hydropower reservoirs in the upper reaches of Yellow River, China.
• Temporal Scale: Long-term operation planning.

Methodology and Data

• Models used: Artificial intelligence models for response surface approximation.
• Data sources: Historical inflow data from cascade reservoirs.

Main Results

• The RSSDP algorithm achieves a significant enhancement in computational efficiency (speedup ratio exceeding 6-fold) compared to traditional stochastic dynamic programming, while maintaining scheduling accuracy comparable to SDP.
• The larger the scale of the cascade system, the more pronounced the efficiency advantage becomes.

Contributions

• This study provides an original contribution by developing an efficient and feasible technical pathway for long-term optimal operation rule formulation for cascade reservoirs using RSSDP.
• The results demonstrate the potential of RSSDP in addressing the "curse of dimensionality" in stochastic dynamic programming, enabling more timely and efficient decision-making in hydropower system operations.

Funding

• This research was funded by the State Key Laboratory of Water Disaster Prevention (Grant No. SKLWDP2022001) and the National Natural Science Foundation of China (Grant No. 52279201).

Citation

@article{Feng2026Artificial,
  author = {Feng, Zhong-kai and Chao, Xuan and Xiao, Yang and Tang, Hong‐Wu and Niu, Wenjing and Zhang, Tao-tao and Kong, Ling-zhong},
  title = {Artificial intelligence-based response surface stochastic dynamic programming for long-term operation rule derivation of cascade hydropower reservoirs},
  journal = {Journal of Hydrology},
  year = {2026},
  doi = {10.1016/j.jhydrol.2026.136216},
  url = {https://doi.org/10.1016/j.jhydrol.2026.136216}
}