On the Developing Reference Frame Transformation for Complex AC Systems from a Data-Driven Perspective

Reference frame transformation plays an important role in multi-phase AC system's control. While most existing ref-erence frame transformations are only applicable to some fixed-phase AC systems, there is a lack of a generalized transformation developing method for complex AC systems whose phase number is very large. In this paper we discuss the reference frame trans-formation for such complex AC systems from the data-driven perspective. Our key idea is based on a physical intuition, saying that a well-designed system is essentially low-rank and sparse. By leveraging this intuition, we perform low-rank decomposition and sparse representation on some high-dimensional AC system data, effectively simplifying them into a low-dimensional DC data. Comparing our transformed output with traditional methods, we find them to be remarkably similar, differing only in their initial angle. This suggests that a reference frame transformation can be developed solely using experimental data without requiring any prior model knowledge, thus providing a model-free solution for complex systems where model knowledge is not available.

参考坐标系变换在多相交流系统的控制中起着重要作用。然而，现有的大多数参考坐标系变换仅适用于某些固定相数的交流系统，对于相数非常大的复杂交流系统，缺乏一种通用的变换推导方法。在本文中，我们从数据驱动的角度讨论了此类复杂交流系统的参考坐标系变换。我们的核心思想基于一种物理直觉，即一个精心设计的系统本质上是低秩且稀疏的。利用这种直觉，我们对一些高维交流系统数据进行低秩分解和稀疏表示，有效地将它们简化为低维直流数据。将我们的变换结果与传统方法进行比较，我们发现它们非常相似，只是初始角度不同。这表明参考坐标系变换可以仅使用实验数据来推导，而无需任何先验模型知识，从而为无法获取模型知识的复杂系统提供了一种无模型解决方案。