分布式优化在互联电网中的应用分析及展望

Application Analysis and Prospect of Distributed Optimization in Interconnected Power Grid

在新能源随机性与波动性的影响下,互联电网能量与灵活性的跨区域互济重要性凸显,有效的分布式优化算法是不可或缺的技术支撑。为了在互联电网不同运行层级的重要场景中选择适宜的算法,文中归纳了分布式优化建模和求解的一般原理,并提出将现有分布式算法分为原始可行方向、对偶可行方向、原对偶、切平面、列生成和多参数法6类。针对互联电网的多区域输电网协同调度、输配协同优化调度以及配电网分布式资源协同调度不同运行层级这3个重要场景,阐明了采用分布式优化的必要性,总结了统一的变量/约束耦合问题的典型建模及变换关系,并分析了6类算法在各场景的实施情况和应用优缺点。从物理场景、数学模型以及算法整体角度分析分布式优化在互联电网3个重要场景应用存在的问题,并给出了不同场景的算法选取建议。结合配电网运行新场景、数学建模新问题、算法设计新技术,展望了未来互联电网应用分布式优化的方向。

Under the influence of randomness and fluctuations of renewable energy,the inter-regional mutual exchange of energy and flexibility of interconnected power grids is of great importance,and an effective distributed optimization algorithm can provide indispensable technical support.In order to select suitable algorithms in the important scenarios on different operation levels of the interconnected power grids,the general principles of distributed optimization modeling and solving are summarized in this paper,and the existing distributed algorithms are divided into 6 types:primal feasible direction,dual feasible direction,primal-dual,cutting plane,column generation and multi-parametric programming method.Aiming at the three important scenarios of coordinated dispatch of multi-regional transmission network,coordinated optimal scheduling of transmission and distribution,and distributed resource coordinated scheduling of distribution network at different operation levels,the necessity of adopting distributed optimization is clarified,the typical modeling and transformation relationship of the unified variable/constraint coupling problem is summarized and advantages and disadvantages in the implementation and application of the six types of methods in each scenario are analyzed.From the perspectives of physical scenarios,mathematical model and overall algorithm,the problems in the application of distributed optimization in three importamt scenarios of interconnected power grids are analyzed,and the algorithm selection suggestions for different scenarios are given.Combined with new scenarios of distribution network operation,new problems of mathematical modeling,and new techniques of algorithm design,the future directions of distributed optimization of interconnected power grid applications are prospected.