计及动态一致性风险度量的水电短期优化调度

Short-term Optimal Regulation of Hydropower Plants With Dynamic Coherent Risk Measures

发电量的多期投标组合是一个动态的优化问题,决策过程中常常呈现多期风险,因而对风险的度量也应该是动态的。条件风险价值(conditional value at risk,CVaR)是一种静态一致性风险度量,不适用于对发电量的时间分解进行多期风险评估。该文提出一种动态一致性风险度量,考虑风险对未来投资收益波动的长期影响,将分位数作用于静态一致性风险度量来表征多期风险的动态特征,并采用分位数回归的方法,以各期CVaR的绝对偏差加权和最小为目标函数建立数学模型,该模型可同时应用于计及风险的发电量时间分解和空间分配计算。以水电厂短期优化调度为例,说明该风险度量指标的可行性和实用性。

The multi-period portfolio strategy is a dynamic optimal problem, i.e., a process of decision-making with multi-period risks, whose risks should be measured by a dynamic index accordingly. The conditional value at risk （CVaR） is a static coherent risk measure, which can not be used to make a multi-period risk evaluation on the time distribution of power generation. The long-term impact of risks on the fluctuation of future yields was considered, the quantile on the static coherent risk measure was acted to reflect the dynamic characteristic of multi-period risks, a novel dynamic coherent risk measure was proposed, and a model was built by introducing the quantile regression to minimize the weighted sum of the absolute deviation of each CVaR. The model can be applied to both the time and space distributions of power generation with risk constraints. A short-term optimal regulation of hydropower is simulated to show that the proposed risk measure is feasible and applicable.