In this brief, a surplus-based approach is proposed to solve the distributed optimal resource allocation problem over directed graphs. Two features distinguish our method from the existing distributed resource allocation schemes. The first one is that the proposed method can be implemented on general strongly connected graph and overcomes the asymmetry caused by the digraphs. The second one is that the proposed method enjoys the privacy-preserving property and avoids the privacy information leakage in the process of information exchange. The distributed algorithm is designed and analyzed by virtue of primal–dual methods, projected gradient, and eigenvalue perturbation theory. Furthermore, distributed economic dispatch in power system, as a simulation example, is used to illustrate the effectiveness of the presented distributed algorithm.

在本文中，提出了一种基于剩余的方法来解决有向图上的分布式最优资源分配问题。有两个功能将我们的方法与现有的分布式资源分配方案区分开来。第一个是所提出的方法可以在一般的强连通图上实现，并克服了有向图引起的不对称性。第二个是提出的方法具有隐私保护性能，避免了信息交换过程中隐私信息的泄露。分布式算法是根据原始对偶方法，投影梯度和特征值微扰理论进行设计和分析的。此外，以电力系统的分布式经济调度为例，说明了该分布式算法的有效性。