Analysis Error Correction Method For Quasi-Synchronous DFT When Signal Frequency Deviated And Its Application

As a common technical method,harmonic analysis and measurement is widely used in power systems and other industrial and agricultural fields,and Discrete Fourier Transform(DFT)is the most widely used harmonic analysis method.However,DFT method cannot suppress long-range leakage caused by rectangular window sampling and short-range leakage caused by asynchronous sampling.Quasi-synchronous DFT harmonic analysis method(QSDFT)combined with quasi-synchronous sampling algorithm and discrete Fourier transform,can significantly suppress long-range leakage,but can not suppress short-range leakage effectively.Therefore,the harmonic analysis result of QSDFT still has a large error when the signal frequency deviated.The quasi-synchronous sampling algorithm performs signal sampling by multiple iterations and complexization integration method,its discrete spectrum has a small main lobe width and fast side lobe attenuation.Therefore,it can effectively suppress long-range leakage and is an ideal sampling window function.However,when the signal frequency deviates,there will still be great harmonic analysis errors caused by short-range leakage.Moreover,the characteristics of QSDFT multiple iterations make its discrete spectral function extremely complex,and the inversion function can not be obtained to deduce the correction function relationship between signal frequency deviation and harmonic error.In order to solve the problem of difficult to correct short-range leakage error of QSDFT,two methods,Variable Picket Fence(VPF)and linear correction algorithm(LCA),are proposed.The main research contents and innovative achievements of this paper are as follows:(1)Aiming at the short-range leakage problem of QSDFT,the idea of variable picket fence sampling in frequency domain is proposed,and the Variable Picket Fence harmonic analysis method(VPF)based on this idea is given.The short-range leakage of QSDFT is caused by the fact that the frequency domain sampling fence at a fixed position cannot be aligned with the harmonic discrete spectrum peak when the signal frequency is deviated.Based on the study of QSDFT spectrum stretching characteristics,the idea of variable picket fence is proposed.And the variable picket fence means that the position of the frequency domain sampling during harmonic analysis is not fixed,but changes according to the deviation of the signal frequency.On this basis,the VPF harmonic analysis method is presented.The simulation results show that VPF can effectively suppress short-range leakage within ±10%signal frequency deviation and obtain high-precision harmonic analysis results.(2)Aiming at the problem that the VPF is large amount of computation and cannot be calculated step by step,a harmonic amplitude linear correction method based on least squares theory and a harmonic phase linear correction method based on linear error theory are proposed,and these two methods constitute the linear correction algorithm of QSDFT(LCA).In the harmonic amplitude linear correction method,the amplitude error curve of QSDFT is constructed by least squares method,and then the correction equation of the amplitude and signal frequency deviation of QSDFT is obtained.The linear error characteristics of the QSDFT harmonic phase show that the harmonic phase error is only linearly related to the signal frequency,harmonic order and iteration number,and then the correction equation of the phase and signal frequency deviation of QSDFT is obtained.The simulation results show that LCA can significantly reduce the short-range leakage when the signal frequency deviated,and improve the harmonic analysis accuracy of QSDFT.At the same time,it can reduce the time complexity and realize real-time analysis.(3)Aiming at the problem of remote measurement of phase difference of power frequency signal,a Distributed Non-synchronous Phase Difference measurement method(DNPD)is proposed.This method employs LCA to suppress long-range and short-range spectral leakage,introduces 220V mains signal as reference signal to realize distributed measurement and non-synchronous measurement of phase difference,and accurately calculates initial phase and phase difference of the signal similar to sinusoidal signal by Newton iterative initial phase calculation method.The simulation experiment and engineering application show that DNPD can accurately measure the phase difference remotely and asynchronously within ±1%frequency deviation,which has high engineering value.(4)Aiming at the problem of resistive current extraction of Metal Oxide Surge Arrester(MOSA),a resistive current extraction method for the online monitoring system of MOSA is proposed,which is called Distributed Non-synchronous Resistive Current measurement method(DNRC).The method employs LCA and DNPD to measure the resistive current parameters of the MOSA remotely and non-synchronously.The method can realize the sampling of the current/voltage signal of MOSA without the long-distance signal line,and also these samplings can be completed not at the same time.DNRC solves the problem of the resistive current extraction of the MOSA ideally.The simulation results and field experiments show that DNRC can accurately measure the different resistive current parameters of MOSA in various complex field environments.