A Simple Method to Localize and Monitor the Emissive Sources in a Converter Panel

Abstract

The converter panels are generally compactly packed with converters, batteries, and controllers. Switching process of the converters generates electromagnetic interference and has the potential to interfere with the regular operation of other parts of the panel as well as the peripheral devices. Localization is the first step for continuous monitoring of the radiation level of these converters. Generally, 2-D methods based on estimation of signal parameters via rotational invariance techniques (ESPRIT) and multiple signal classification (MUSIC) are used to localize these emissive sources. However, the high efficiency of these methods comes with the cost of a burden. Algorithms that combine 1-D ESPRIT and 1-D MUSIC reduces the 2-D complexity, but with the less number of the snapshot and low signal to noise ratio (SNR), the performance of ESPRIT reduces. In this article, a modified two-stage MUSIC (MTS-MUSIC) algorithm is proposed to overcome the aforementioned limitations. MTS-MUSIC utilizes the advantages of MUSIC for estimating both angle of arrival and the range of the sources. This method performs better than the ESPRIT–MUSIC combination for less number of snapshots and low SNR. However, for closely spaced sources, even proposed MTS-MUSIC shows poor resolution and some time fails to identify the sources. The proposed method further combines spatial downsampling with MTS-MUSIC to address this scenario. The MTS- MUSIC algorithm reduces the computational burden, and spatial downsampling enhances the resolution. A set of near-field source localization experiments are conducted, and the improved resolution of experimental results validate the proposed method.