基于压缩感知估计的GIS击穿放电定位方法研究*

经典的空间谱估计方法，如多重信号分离(MUSIC)方法，对噪声敏感，难以在低信噪比环境中有效地进行波达方向估计。为提升在复杂电力环境中的气体绝缘金属封闭开关设备(GIS)击穿定位能力，该文提出了一种基于极化内插的压缩感知波达方向估计方法 CSP-DOA。该方法对传声器阵列接收到的数据进行建模，形成多测量矢量模型，结合压缩感知中稀疏重构技术进行波达方向估计；同时，该方法还采用极化内插技术解决了稀疏重构中的费网格问题，进一步提升了波达方向估计精度及计算效率。通过数值模拟对算法的定位效果进行了分析，仿真结果表明CSP-DOA方法对于击穿信号有更好的定位效果。结合可见光图像匹配实现了GIS击穿信源的二维可视化定位，在某高压大厅的GIS模型上进行耐压击穿定位试验研究，试验结果进一步验证了该文方法可较好的应用于GIS的击穿定位。

Research on GIS breakdown discharge location method based on compressed sensing estimation

Classical spatial spectrum estimation methods, such as music, are sensitive to noise and difficult to estimate DOA effectively in low SNR environment. In order to improve the ability of GIS breakdown location in complex power environment, this paper proposes a spatial spectrum estimation method csp-doa based on optimized compressed sensing technology. In this method, the data received by the microphone array is modeled to form a multi measurement vector model, and the direction of arrival (DOA) is estimated by combining sparse reconstruction technology in compressed sensing; At the same time, polarization interpolation technology is used to solve the problem of sparsity grid in sparse reconstruction, which further improves the accuracy and efficiency of DOA estimation. The simulation results show that csp-doa method has better positioning effect for breakdown signal. Combined with the visible light image matching, the two-dimensional visual location of GIS breakdown source is realized. The experimental research on breakdown location is carried out on the GIS model of a high voltage hall. The experimental results further verify that the method can be better applied to the breakdown location of GIS.