基于风电出力模糊集的电-气耦合系统分布鲁棒优化调度

随着可再生能源装机容量不断扩展及综合能源系统研究的持续推进,燃气轮机作为耦合元件加深了电力系统和天然气系统的相互融合,为可再生能源消纳提供了新的途径。针对随机规划方法、区间优化及鲁棒优化算法处理风电出力不确定性的不足,文中提出一种基于风电预测误差模糊集的分布鲁棒优化方法求解电力-天然气耦合系统的协同调度决策。首先,利用主成分分析法提取高维预测误差向量蕴含的时空尺度关联特性,并引入一系列矩函数描述预测误差的分布信息以构建相应的模糊集。然后,建立两阶段分布鲁棒优化经济调度模型,第1阶段制定日前机组开停机计划、调度出力及备用方案;第2阶段辨识最劣风电场景分布以保证第1阶段调度决策的有效性。结合线性决策规则和对偶理论将该半无限优化问题转化为有限维优化问题进行求解。最后,通过算例验证了所提模型及求解方法的有效性。

Distributionally Robust Optimal Dispatch for Power-Gas Coupled System Based on Fuzzy Set of Wind Power Output

With the continuous expansion of renewable energy installation capacity and the persistent advancement of integrated energy system research, the gas-fired units deepen the integration of power system and natural gas system as the coupled components and provide a new way for renewable energy accommodation. Since the stochastic scheduling method, interval optimization and robust optimization algorithms have their own shortcomings in dealing with wind power uncertainty, a distributionally robust optimization approach based on fuzzy set of wind power prediction error is proposed to solve the collaborative dispatch problem for power-gas coupled system. Firstly, the principal component analysis is adopted to extract the temporal-spatial scale correlation characteristics of high-dimensional prediction error vector, and a series of moment functions are introduced to describe the distribution information of prediction errors to construct the corresponding fuzzy set. Then, a two-stage distributionally robust optimal ecomomic dispatch model is established. The first stage is used to make the unit commitment decision, power output and reserve configuration by units, and the second stage is used to identify the worst-case distribution of wind power to ensure the effectiveness of the scheduling strategy determined in the first stage. Combining the linear decision rule and dual theory, the above semi-infinite optimization problem can be transformed into a finite-dimensional optimization problem and then solved. Finally, the simulation results verify the validity of the proposed model and solution method.