A Distributed Double-Newton Descent Algorithm for Cooperative Energy Management of Multiple Energy Bodies in Energy Internet
A Distributed Double-Newton Descent Algorithm for Cooperative Energy Management of Multiple Energy Bodies in Energy Internet
复制标题
DOI：
10.1109/tii.2020.3029974
复制
发表时间：
2020-10
影响因子：
12.3
通讯作者：
Yushuai Li;D. Gao;Wei Gao;Huaguang Zhang;Jianguo Zhou
中科院分区：
文献类型：
--
作者：
Yushuai Li;D. Gao;Wei Gao;Huaguang Zhang;Jianguo Zhou
本文研究了多能量体分布式协同能量管理问题，既考虑单个能量体内各参与者的最优发电量/能耗，又考虑任意对能量体之间互联线上的最优能量分配。首先，我们定义了由许多能量体组成的系统的物理和通信结构，每个能量体都被视为多能产消者。然后，提出了一种分布式能源管理模型，既能实现能源生产和消费的整体利润最大化，又能实现能源输送的成本最小化。针对这一问题，提出了一种分布式双牛顿下降（DDND）算法，该算法具有两个优点。一方面，通过利用二阶信息，将牛顿下降的概念嵌入到算法的实现中，提高了算法的收敛速度；另一方面，所提出的算法以完全分布式的方式执行。因此，各参与者既能在本地获得其最优运行，又能获得全球能源市场出清价格；同时，每个能量路由器都能在本地获得与相邻能量路由器交换的最优能量。此外，我们还证明了所提出的DDND算法能够渐近收敛到全局最优点。从理论上保证了DDND算法的正确性。最后，仿真结果验证了该算法的有效性。
This article investigates the problem of distributed cooperative energy management of multiple energy bodies with the consideration of both the optimal energy generation/consumption of each participant within single energy body and the optimal energy distribution on the interconnected lines between any pair of energy bodies. First, we define the physical and communication structure of the system formed by many energy bodies, each of which is viewed as a multienergy prosumer. Then, a distributed energy management model is proposed to achieve not only maximum profits of overall energy generation and consumption, but also minimum cost of energy delivery. To address this issue, a distributed double-Newton descent (DDND) algorithm is proposed, which possesses two advantages. On the one hand, by employing second-order information, the concept of Newton descent is embedded into the implementation of the proposed algorithm, resulting in faster convergence speed. On the other hand, the proposed algorithm performs in a fully distributed fashion. As a consequence, each participant can locally obtain its optimal operation as well as the global energy market clearing prices; meanwhile, each energy router can locally obtain the optimal exchanged energy with its neighbor energy routers. Moreover, we prove that the proposed DDND algorithm can asymptotically converge to the global optimal point. As a result, the correctness of the DDND algorithm can be guaranteed in theory. Finally, simulation results validate the effectiveness of the proposed algorithm.