Harmonics Detection And Analysis Based On Adaptive Algorithm In Power System

With the rapid development of power electronics technology, more and morenon-linear devices are used widely in various industries. Many serious problems aremade because the harmonics generated by the equipment inject into the power system.So how to control harmonics becomes the issue to be concerned about. Among them,the first condition for harmonic elimination is to get an accurate harmonic parameterestimation values. Moreover, the harmonic is very complex in power system, so it’snecessary to search the suitable harmonic analysis methods for the differentharmonics objects.Nowadays, Windowed FFT is used widely in the steady-state harmonic detection.But there are several problems, such as leakage and barrier effect, which influence theaccuracy of harmonic parameter estimation values. To solve those problems, twoneighboring spectral interpolated harmonic algorithm which can modify signalestimated value first is analyzed. The simulation analysis of hanning windowinterpolated harmonics is carried with a steady-state signal whose highest orderharmonic is25, the results show that two neighboring spectral interpolated harmonicalgorithm is an accurate steady-state harmonic analysis method.Secondly, an adaptive harmonic analysis method based on least mean squarealgorithm is analyzed. The adaptive algorithm transforms harmonic parameterestimation problem into solving process of the weight vector which is throughcontinuous iteration to close the true solution. And the algorithm uses the feedback ofinput signal and error signal to adjust the iterative process of the weight vectorw n at the adaptive system. Then, the signal parameter estimates are obtained by theconverged weight vector after the iterative process. Through the analysis of the steadystate harmonic, time varying harmonic and adding noise signal, the simulation resultsprove that at static state, the algorithm compared with the FFT algorithm has a highfrequency detection precision, and the accuracy of the parameter estimate increaseswith the convergence factor. However, the convergence of the algorithm is slow, and itis necessary to increase the simulation time in order to get the same accuracy of theparameter estimates. Under time-varying signal, the algorithm has a certain trackingperformance and the parameter estimation accuracy of the algorithm increases withthe convergence factor. Under adding noise signal, the parameter estimation accuracyof the algorithm increases with the Signal to Noise Ratio.Finally, an adaptive harmonic analysis method based on recursive least square algorithm is analyzed. Compared with the previous algorithms, the algorithm has thefaster convergence and the higher ability to track time-varying signals. The simulationresults prove that the algorithm is a high-precision frequency detection method. Withthe increasing of the forgetting factor, the parameter estimation accuracy reduces. Instatic state, the parameter estimation accuracy of the algorithm is higher than the FFTalgorithm, and the convergence rate is much faster than the least mean squarealgorithm. Under time-varying signal, the tracking speed increases with the forgettingfactor, and the tracking ability of the algorithm is higher than the previous methods. Inthe case of adding noise, the algorithm is greatly affected by the noise. Simulationresults show that the algorithm has some practical value.So it can be used in real-timeonline detecting of signal.