Optimal Stochastic Tracking for Primary Frequency Control in an Interactive Smart Grid Infrastructure
Optimal Stochastic Tracking for Primary Frequency Control in an Interactive Smart Grid Infrastructure
复制标题
交互式智能电网基础设施中一次频率控制的最优随机跟踪
作者：
Ding Li;S. Jayaweera
提出了一种用于交互式智能电网基础设施的最优随机跟踪方案，该方案由三个步骤组成：1）电力公司通过对所请求的客户负荷进行需求响应来重塑客户负荷分布; 2）个人智能家庭做出最优的购电顺序决策; 3）有功功率平衡的最优随机控制方案（一次频率控制）的设计，在客户负载和分布式可再生能源发电产生的不确定性的存在。与前两部分在我们以前的工作中解决，本文中，我们专注于一次频率控制方案的设计，在多层控制架构，以稳定频率和维持分布区域内的有功功率平衡。我们提出了两种基于同步发电机状态空间表示的随机跟踪方案：1）基于参考动态的跟踪和2）基于参考统计的跟踪。我们进一步扩展建议的最优控制器，考虑来自不同客户的异步负载信号的现实情况。为了补偿不同的客户信号所看到的不同延迟，提出了一种基于卡尔曼滤波器的预测方案来估计正确的参考信号。我们表明，集中式参考预测可以等效地实现分布式。仿真结果表明，所提出的预测和跟踪计划的性能。
An optimal stochastic tracking scheme is proposed for an interactive smart grid infrastructure made of three steps: 1) The utility reshapes the customer load profiles by scheduling a demand response for the requested customer loads; 2) individual smart home makes optimal sequential decisions on power purchase; and 3) optimal stochastic control schemes for the active power balance (primary frequency control) are designed, in the presence of uncertainties arising from customer loads and distributed renewable generations. With the first two parts addressed in our previous work, in this paper, we focus on the primary frequency control scheme design in the multilayer control architecture to stabilize frequency and maintain the active power balance within the distributed areas. We propose two stochastic tracking schemes based on the state-space representation of a synchronous generator: 1) reference-dynamics-based tracking and 2) reference-statistics-based tracking. We further extend the proposed optimal controllers by considering the realistic scenario of asynchronous load signals from different customers. To compensate for different delays seen by different customer signals, a Kalman-filter-based prediction scheme is proposed to estimate the correct reference signal. We show that the centralized reference prediction can equivalently be implemented distributively. Simulation results are presented, showing the performances of the proposed prediction and tracking schemes.