This article proposes an improved sparse signal-based fault location method for distribution networks (DNs), which utilizes only a limited voltage measurements and overcomes the challenge of insufficient measurements in DNs. It addresses the challenge that location accuracy relies heavily on accurate reconstruction and is easily affected by fault position. A sparse overcomplete representation (OCR) recovery model is proposed to capture the fault point by inserting a compact set of virtual buses and finding sufficiently large atoms instead of accurate value reconstruction. Furthermore, considering the large-scale characteristic of the OCR recovery model, the FastLaplace algorithm with Laplace prior is employed to achieve higher sparsity and fast reconstruction rather than Gaussian prior. The proposed method performs high accuracy and robust properties through simulations on the modified IEEE 34 Bus Test Feeder, achieving accurate and efficient fault location compared to the other five algorithms.

中文翻译：

---

配电网稀疏过完备表示故障定位模型及使用 FastLaplace 贝叶斯的高效解决方案

本文提出了一种改进的基于稀疏信号的配电网（DN）故障定位方法，该方法仅利用有限的电压测量，克服了配电网测量不足的挑战。它解决了定位精度很大程度上依赖于精确重建且容易受断层位置影响的挑战。提出了一种稀疏过完备表示（OCR）恢复模型，通过插入一组紧凑的虚拟总线并找到足够大的原子而不是精确的值重建来捕获故障点。此外，考虑到OCR恢复模型的大规模特性，采用拉普拉斯先验的FastLaplace算法比高斯先验获得更高的稀疏性和快速重建。该方法通过对改进的 IEEE 34 总线测试馈线进行仿真，具有高精度和鲁棒性，与其他五种算法相比，实现了准确、高效的故障定位。