面向多主体隐私保护的源网荷储分布式协同优化调度

Privacy-Preserving Multi-agent Coordinated Economic Dispatch with Generation-Grid-Load-Storage Interaction

如何在满足各主体隐私需求的前提下,促进电力灵活配置与可靠供应,是构建隐私安全型多主体互动新型电力系统的核心。为此,本文提出了一种基于改进共识交替方向乘子法(C–ADMM)的分布式协调优化调度方案,仅通过交换相邻主体的耦合支路信息来统筹全局优化,实现协同优化的同时保护了互动主体的隐私数据。首先,针对源网荷储参与日前调度时的主体隐私需求,建立多主体协同调度模型,各相邻主体仅通过共享边界耦合支路阻抗信息实现解耦,各主体的优化调度问题在本区域独立求解,并在迭代过程中不断更新耦合乘子从而实现全局优化。然后,考虑主体中离散型可调负荷对C–ADMM收敛性能的影响,使用割平面法进行变量松弛与整数解逼近,通过移除整数限制和增加新约束的方法实现问题的转化,以获得可调负荷与源网储的协调策略。最后,使用IEEE30节点系统进行仿真计算,具体地,将IEEE30节点系统划分为了不同的主体,分别进行算例测试,验证了所提分布式协同优化调度策略的有效性。实验结果表明:在不同划分方式下,与集中式方法相比,改进C–ADMM算法均能产生相对稳定的收敛解;在不同计算参数下,基于改进C–ADMM与集中式方法的总成本基本相同,最大误差不超过0.42%。同时,与传统C–ADMM方法相比,改进C–ADMM的收敛性指标数量级缩小至百分之一。

Under the prerequisite that the privacy protection requirements of each involved agent are satisfied,ensuring a dependable power sup-ply and flexible power allocation stands as the cornerstone for the privacy-preserving operation of a power system,taking into account the interac-tion of multiple agents.In this paper,we present a distributed economic dispatch scheme utilizing the modified consensus alternating direction method of multipliers(C-ADMM).Global optima are attained by solely exchanging coupling line information among interconnected areas,thereby achieving simultaneous privacy protection and dispatch goals.Firstly,based on the privacy requirements of agents participating in the day-ahead dispatch,we construct a multi-agent collaborative scheduling model.Each adjacent subject achieves decoupling by solely sharing the boundary coupling branch impedance information.The optimal scheduling problem of each subject is independently solved in each area,and the coupled multipliers are continuously updated in the iterative process to achieve global optimization.Considering the impact of distributed ad-justable loads on the convergence of C-ADMM,we employ the cutting-plane method to relax and approximate the integer variables.This facilit-ates the transformation of the problem by removing integer restrictions and introducing new constraints,thereby obtaining the optimal coordin-ated dispatch results for discrete adjustable loads and other resources.Finally,we conduct simulation calculations using the IEEE 30-bus system.Specifically,we partition the IEEE 30-bus system into different subjects and perform case tests to validate the effectiveness of the proposed dis-tributed collaborative optimization scheduling strategy.The results indicate that,under different division methods,the improved C-ADMM al-gorithm can yield relatively stable convergent solutions compared to the centralized method.Under different calculation parameters,the proposed and centralized methods,considering various parameters,are approximately the same,with a maximal error of 0.42%.Additionally,the conver-gence criterion is reduced to one hundredth compared to the conventional C-ADMM.