为解决现有场景生成研究较少关注高维变量与极端场景的问题，本文提出一种水风光多能互补系统的高维耦合场景生成方法。首先，描述径流、风速、光照强度和温度的相关性；其次，基于资源时空相关性矩阵，采用多序列一阶自回归模型模拟得各类资源的高维耦合基础场景集；而后，通过K-means聚类算法提取典型场景；最后，以资源总发电潜力不利为原则，设置不同资源的权重，从而得到不同类型的极端场景。以某水风光互补系统为实例，结果表明：径流与风速之间存在互补性，风光资源呈正相关；模拟的径流、风速、光照强度和温度序列保持了实测序列的统计特征与时空相关性，所得典型场景和极端场景可为极端水文气象条件下水风光互补系统的运行管理提供决策参考。

The carbon neutrality strategy has driven a profound decarbonization transformation of China's energy structure, with an increasing penetration of renewable energies such as wind power and photovoltaic (PV) in the energy system. However, new energy sources such as wind and PV have significant stochasticity, intermittency, and volatility, which can lead to serious curtailment problems when they are integrated into the power grid at a large scale. Leveraging the regulation capability of hydropower to establish hydro-wind-PV complementary energy systems (HWPCES) is a key solution to these challenges. However, HWPCESs are influenced by more complex hydro-logical and meteorological factors, presenting significant challenges for its planning and operation. Therefore, a critical issue that needs to be addressed is how to consider the uncertainty and spatiotemporal correlation of multiple resources to generate representative and extreme high-dimensional coupled scenarios. The study focuses on the generation of high-dimensional coupled and extreme scenarios for HWPCES. First, based on the Pearson correlation analysis, the correlations between runoff, wind speed, solar radiation intensity, and temperature in the HWPCES are revealed. Second, considering the temporal autocorrelation and spatial cross-correlation of each variable, a multi-sequence first-order autoregressive model is constructed to conduct random simulations of high-dimensional coupled scenarios for runoff, wind speed, solar radiation intensity, and temperature. The simulated scenarios are compared with the observed sequences in terms of statistical characteristics and correlations to verify the validity of the generated basic scenario set. Then, representative scenarios that capture the essential variability and trends were derived using the K-means clustering algorithm which could reduce basic scenarios. Finally, based on the principle of adverse total power generation potential, different resource weights are set to consider extreme scenarios with various combinations, and extreme scenarios with different return periods are determined through frequency analysis of the total power generation potential. One HWPCES was selected as the case study. The result showed that there is a significant negative correlation between runoff and wind speed, a positive correlation between runoff and temperature, and an absence of notable complementarity between wind speed and solar radiation intensity. The simulated sequences for runoff, wind speed, solar radiation intensity, and temperature effectively maintained the statistical characteristics and spatiotemporal correlations observed in the measured data. The representative scenarios preserved the seasonal and variability characteristics of each resource. Extreme scenarios of single-resource with different return periods were generated. By assigning resource weights based on the installed capacity of each power station, compound extreme scenarios with high and low total power generation potential were obtained. These scenarios successfully captured the characteristics and overall trends of the observed extreme scenarios. Additionally, the complementarity between runoff, wind, and solar resources maintains a relative balance, preventing simultaneous extremes in all resources. The method proposed in this study addresses the limitations of existing scenario generation research, which primarily focuses on individual runoff or combined wind-solar resources and neglects extreme scenarios. The generated basic scenario set provides data support for the optimization operation of the complementary system, helping to develop operation strategies for different types of scenarios and enhancing the adaptability of the HWPCES. In addition, the generated scenarios can assist in the evaluation of planning schemes by analyzing the output characteristics and power supply reliability of the system under both conventional and extreme hydro-meteorological conditions, leading to the determination of more climate-resilient capacity configuration schemes. Therefore, this method offers technical support for the planning, design, and operation of HWPCESs under extreme hydrometeorological conditions.

国家自然科学基金(U23B20141)；湖北省重点研发计划项目(2022AAA007)；中国长江电力股份有限公司项目(Z242302011)