Analysis Of Harmonic Signal Multifractal Feature Based On Improved Mfdfa

The accurate detection and analysis of harmonics is the precondition for improving power quality and effectively managing power system.As the time-domain waveform and changing trend of harmonics is featured with self-similarity known as fractal feature,and a single fractal cannot describe the local fractal characteristics of the signals,the multifractal theory is applied in harmonic analysis.Based on the excellent stability and high accuracy of multifractal detrended fluctuation analysis(MFDFA),an improved MFDFA algorithm combined with empirical mode decomposition(EMD)theory and sliding windows technology is proposed to overcome some shortcomings of the traditional algorithm in the process of identifying the detrended function in detrended fluctuation.Targeting at the non-linear and non-stationary characteristics of harmonics,and the essential similarities in the signal processing between wavelet and fractal techniques,wavelet theory is combined,and even harmonics(including second harmonics,and fourth harmonics)and odd harmonics(including third harmonics,fifth harmonics,and eleventh harmonics)are selected to analyze the multifractal feature of signals.Finally,the relationship between multifractal characteristics of harmonic signals and harmonic frequencies is concluded.Firstly,the original signal is decomposed according to the multi-resolution analysis characteristics of wavelet,and the components of different frequency bands are obtained.Subsequently,the signal components are simulated by traditional algorithm and improved algorithm respectively,and the multifractal characteristic components extracted from the multifractal spectrum graph,the q-order Hurst exponent graph,the q-order quality index graph and the spectrum graph are analyzed.As the results reveal,the multifractal spectrum and its parameters can perfectly characterize the multifractal characteristics of harmonic signals.Harmonic signal presents obvious multifractal characteristics,which is strengthened with the increase of signal frequency.Compared with the traditional algorithm,the improved MFDFA algorithm can more effectively overcome the shortcomings of the traditional one while better satisfying the detrended requirements.Besides,the curve fitting becomes more stable,and the accuracy of multifractal feature extraction is also improved.Therefore,these results are of great significance to the identification of harmonic category and the targeted governance for harmonics in the next step.