A Study Of Power Measurement Of Nonlinear Load Based On Discrete Wavelet Transforms

In the modern electric power systems, the proliferation of nonlinear loads such aselectric arc furnaces (EAF), electric locomotives, tremendously led to the high harmonicdistortion level by injecting harmonics into the grid. These harmonics include harmonics,inter-harmonics and sub-harmonics. Furthermore, when the distorted waveform changestimely, Fourier transforms of such waveform no longer concentrate at discrete frequencies,and the energy associated with each harmonic component occupies a particular regionwithin the frequency band. In such cases, traditional power measurement methods based onFFT is inadequate.This thesis takes the study of theory and related algorithm for power measurement ofnonlinear loads as the goal, focusing on research of several metering algorithm of relatedpower quantities. The achievements include:(1) studied the electrical characteristics ofnonlinear load, and discussed the limitations of the present FFT-based algorithm;(2)Analyzed the advantages and disadvantages of most commonly used harmonic analysisalgorithms–fast Fourier transform (hereinafter referred to as FFT) based and wavelet packetdecomposition based ones, and proposed a new composite approach. Simulation results andfield tests have proven that more accurate harmonic phase angle can be obtained base on theproposed method compared with traditional FFT method, when the test signal is heavilydistorted by noise. For example, measurement errors of harmonics phase are less than0.2degrees when the signal is composed of15%odd and3.75%even harmonics,1%0.4th and0.8th subharmonics;(3) considering the continuous Fourier spectrum of nonlinear loads,proposed a new discrete wavelet packet based method for measurement of power quantitiesdefined in the standard IEEE1459-2000. This approach gives the concept of the concepts ofthe minimum frequency band width and the corresponding number of decomposition layer,in order to reduce the effects of inter-harmonics. Simulation results of different types ofloads have proven high measurement accuracy of this approach. For example, for the load ofelectric arc furnace, which is rich in inter-harmonics, measurement error of fundamentalactive power is less than0.05%;(4) proposed a new power system frequency estimationmethod, to achieve the amended version of the before-mentioned DWPT method when thepower system frequency fluctuates. The proposed algorithm is verified by static, dynamictests and field measurement. It can achieve more accurate estimation compared to the present methods, when the test signal is heavily distorted by harmonics, inter-harmonics andwhite noise, and measurement error is less than25mHz.