The Research And Implementation Of Inter-harmonic Detection Methods In Power System

Nowadays, the Inter-harmonic is one of the power quality problems which causedparticular concern for people after harmonic. The Inter-harmonic is a special harmonic, and itssources and harms exist extensively, and has the characteristic of non-linear, randomness,distribution, instability and so on. Detecting the inter-harmonic components accurately is thefirst step of effective analysis, compensation, suppression to it, so it is very important tomeasure the inter-harmonic.The frequency, amplitude and phase of inter-harmonic can be obtained accuratelythrough combine modified hamming Burg and RSVM algorithm in this thesis, and thecomputation quantity is small, this afford good basic for effective analysis, compensation,suppression to it. Firstly, the sources, harms, and limits of inter-harmonic are introduced, andthe basic situation of the inter-harmonic research between the domestic and foreign aresummarized in this paper. Secondly, spectrum estimation based on auto-regressive model isdiscussed among classic spectrum and modern spectrum, and the major methods of solve theAR model parameters are introduced. The simulating results demonstrate that Burg algorithmis the most simple. Moreover, as for the peak bias, spectral line splitting, peak disappear, themodified hamming weighted burg algorithm for detect the frequent components are presented.Thirdly, robust support vector machine algorithm is introduced to detect amplitude and phaseof inter-harmonic, building inter-harmonic model with RSVM algorithm and each frequencyof detected was corrected by RSVM which could get accurate frequency, the amplitude andphase of inter-harmonic were obtained. Finally, voltage fluctuation and flicker caused bysignal inter-harmonic and inter-harmonic pairs are analysed with modified hamming weightedBurg and RSVM.Simulating results demonstrate that when detecting the frequency component with themodified hamming weighted Burg algorithm, the peak shifting is reduced, spectral splitting isnot easy to appear, and the phenomenon of spectral disappear is not occurred when low signalto noise ration, the accuracy of frequency detection is improved; Combine modified hammingBurg and RSVM algorithm can detect inter-harmonic component accurately and thecomputation quantity is small; Modified algorithm can detect flicker signal informationaccurately.