This study addressed the problem of multi‐area state estimation in a clustered distribution system. Distribution networks are inherently expansive and comprise a multitude of nodes. This issue increases the state estimation computation time and makes it inapplicable for control of sophisticated distribution networks. Multi‐area state estimation is a technique to reduce computation time while concerning computation accuracy. Many efforts are required to reach a perfect algorithm, followed by the optimization of different parameters of the proposed algorithms. This paper performed a precise mathematical analysis of the impact made by the common (shared) node exchanged information between the areas in the multi‐area state estimation algorithm. Furthermore, a new iterative multi‐area state estimation algorithm equipped with machine learning tools was designed based on analytical detections for enhancing the convergence speed and accuracy of the estimation results. The improvement was evaluated in two clustered networks with 356 and 711 nodes. The results indicated the benefits provided by the proposed modification in terms of convergence speed and accuracy with minimum data exchanges in an iterative multi‐area distribution network.

中文翻译：

---

使用卡尔曼滤波器估计的Takagi-Sugeno模型在配电网中进行多区域状态估计

这项研究解决了集群分布系统中的多区域状态估计问题。配电网络具有固有的扩展性，并包含多个节点。此问题增加了状态估计的计算时间，并使其不适用于复杂的配电网络。多区域状态估计是一种在降低计算时间的同时又提高了计算精度的技术。为了达到理想的算法，需要付出很多努力，然后才能优化所提出算法的不同参数。本文对多区域状态估计算法中区域之间公共（共享）节点交换信息所产生的影响进行了精确的数学分析。此外，在分析检测的基础上，设计了一种新的具有机器学习工具的迭代多区域状态估计算法，以提高估计结果的收敛速度和准确性。在具有356个和711个节点的两个群集网络中评估了这种改进。结果表明，在迭代的多区域分布网络中，在最小化数据交换的情况下，所建议的修改在收敛速度和准确性方面提供了好处。