A novel method on harmonics detection based on compressive sampling matching pursuit

The decomposition process for the extraction of harmonics components from compressed power quality data was required under the frame of Shannon sample theory, but it increased the complexity of data procedure. A compressed sampling matching pursuit (CoSaMP) method was presented to detect harmonics from compressive sensing (CS)-based compressed power data sequence avoiding decompression pretreatment. First, the compressed power data sequence was obtained from original power signal under the manipulation of a measurement matrix based on CS theory. A vector served as a proxy for the original power signal was formed because of energy approximation between them. The largest harmonics component of the proxy was located and extracted using a least-square algorithm. Then, the proxy was updated to reflect the current residual and was used to estimate the next largest harmonics component. This process was repeated until all of the harmonics components were recovered. The CoSaMP harmonic detection method can detect harmonics components from compressed power quality data directly. It has great potential to reduce the burden of sampling devices and save intermediate storage space. Simulation results show the proposed method is effective for both harmonic and inter-harmonic detection in power system.