含CVaR及增广ε-约束法的多目标光储充电站容量优化配置

Optimal Allocation of Multi-Objective Photovoltaic Energy Storage Charging Station Capacity with CVaR and Augmented ε-Constraint Method

电动汽车充电需求和光伏出力的不确定性使光储充电站的收益存在一定的不确定性。为了量化该不确定性给光储充电站收益带来的影响,即收益的CVaR值,建立了含收益和CVaR的光储充电站多目标容量优化配置模型。模型中CVaR函数为条件风险价值量化的投资商的预期收益,收益期望值函数为考虑充电需求和光伏出力典型场景下的收益与其概率乘积之和,结合增广ε-约束法以收益为主目标,将预期收益CVaR次要目标作为约束。求解模型获得不同风险偏好下收益期望值和预期收益CVaR的pareto前沿以及对应的配置容量,采用熵权-TOPSIS法筛选出客观决策方案。相较于传统风险管理中的线性加权法,增广ε-约束法处理目标后可得到更具分布性以及边界最优性的前沿,能够映射出多目标问题的实际pareto前沿,提供收益期望值和预期收益CVaR划分更细致的投资方案。

The uncertainty of electric vehicle charging demand and photovoltaic output leads to a certain degree of uncertainty in the revenue of photovoltaic energy storage charging stations.In order to quantify the impact of this uncertainty on the revenue of photovoltaic energy storage charging stations,namely the CVaR value of revenue,a multi-objective capacity optimization allocation model for of photovoltaic energy storage charging stations is established,including revenue and CVaR.The CVaR function in the model is the expected revenue of the investor quantified by conditional value at risk,and the expected revenue function is the sum of the revenue and its probability product under typical scenarios of charging demand and photovoltaic output,combined with augmenta⁃tionε-constraint method focuses on revenue as the main objective,with the expected revenue CVaR as the secondary objective as the constraint.The pareto frontier of expected revenue and expected revenue CVaR under different risk preferences are obtained by the solution model,as well as the corresponding allocation capacity,and the entropy weight TOPSIS method is used to screen out objec⁃tive decision-making solutions.Compared with the linear weighting method in traditional risk management,the augumentedε-constraint method can obtain more distributed and boundary optimal frontiers after processing the objectives,which can map the actual pareto frontiers of multi-objective problems and provide more detailed investment schemes for dividing the expected revenue value and expected revenue CVaR.