论文标题

部分可观测时空混沌系统的无模型预测

Droop-e: Exponential Droop as a Function of Power Output for Grid-Forming Inverters with Autonomous Power Sharing

论文作者

Kenyon, Rick Wallace, Sajadi, Amirhossein, Hodge, Bri-Mathias

论文摘要

储层计算是预测湍流的有力工具,其简单的架构具有处理大型系统的计算效率。然而,其实现通常需要完整的状态向量测量和系统非线性知识。我们使用非线性投影函数将系统测量扩展到高维空间,然后将其输入到储层中以获得预测。我们展示了这种储层计算网络在时空混沌系统上的应用,该系统模拟了湍流的若干特征。我们表明,使用径向基函数作为非线性投影器,即使只有部分观测并且不知道控制方程,也能稳健地捕捉复杂的系统非线性。最后,我们表明,当测量稀疏、不完整且带有噪声,甚至控制方程变得不准确时,我们的网络仍然可以产生相当准确的预测,从而为实际湍流系统的无模型预测铺平了道路。

This paper presents the novel Droop-e grid-forming inverter control strategy, which establishes an active power-frequency relationship based on an exponential function of the inverter power dispatch. The advantages of this control strategy include an increased utilization of available headroom, mitigated system frequency dynamics, and a natural limiting behavior, all of which are directly compared to the hitherto standard static droop approach. First, the small signal stability of the Droop-e control is assessed on a 3-bus system and found stable across all possible inverter power dispatches. Then, time-domain simulations show improved frequency dynamics at lower power dispatches, and a limiting behavior at higher dispatches. Finally, a novel secondary control scheme is introduced that achieves power sharing following the primary Droop-e response to load perturbations, which is shown to be effective in time-domain simulations of the 3- and 9-bus systems; comparative simulations with a static 5% droop yields unacceptable frequency deviations, highlighting the superiority of the Droop-e control.

扫码加入交流群

加入微信交流群

微信交流群二维码

扫码加入学术交流群,获取更多资源