Research On Characteristic Analysis And Data Compression Method Of Insulator Leakage Current

The transmission line is the main part of power system, and the insulator plays animportant role in supporting the transmission and isolation. Insulator fault may lead tothe accident of power network，thus the on-line monitoring of insulator is verysignificant. Any insulator fault can lead to the change of the insulator leakage current,which is an important parameter to characterize the running state of transmission line.Therefore, the leakage current method becomes the main method of on-linemonitoring of transmission line.In order to find abnormal discharging of insulator, the sampling frequency ofleakage current should be relatively high, which leads to a large amount of data. Lowrates wireless network is used for data transmission of leakage current. Although thereare kinds of methods to make the data compression of the insulator leakage currentdata, each one has its limitations. Making data compression and reconstruction ofleakage current with the compressing sensing theory can achieve high reconstructionaccuracy with high compression ratio. However, the direct compression of the originalleakage current data would lead to the loss of pulses.This paper analyses the frequency spectrum and the power spectrum of theinsulator leakage current, achieves the characteristic of periodicity and non-stability.Then this paper classifies the insulator leakage current according to the time domaincharacteristic and makes the separation of the insulator leakage current to pulse andnon-pulse, and then compresses the data of the pulses with low compression ratio andthe data of the others with high compression using compressed sensing theory, makesthe combination during the reconstruction. The reconstruction error is reduced greatlyin case of similar compression ratio and also retaining the pulses.Determination of measurement matrix is the significant step for realizing thecompressing sensing theory. The compression and recovery effect of classicmeasurement matrix is already good, but it can still be further optimized. Pointing atthe periodicity and unsteady characteristic of leakage current, the paper comparesmultiple measurement matrix of their effect via experiments, putting forward to makedata compression and reconstruction of leakage current using Toeplitz matrix,circulant matrix and sparse matrix as measurement matrix, of which the reconstitutioneffect is almost the same as classical measurement matrix and depletes computationalcomplexity and workload.