The Research And Analysis Of Harmonic Detection Based On HHT

With the technological level of social progress and constantly improved the power electronics technology, a variety of impact resistance, non-linear loads are widely used in industry.Thus, the distribution network to highlight the growing problem of harmonic pollution, power quality problem is getting worse. So, on an appropriate harmonic detection method, the parameters for real-time power quality, accurate detection and analysis, providing the basis for taking effective measures for harmonic suppression and control, has a very important practical significance.Firstly, this thesis introduces several traditional harmonic detection method. The first harmonic measurement is achieved using analog filters, because the error is too large, real bad, harmonic measurement process is no longer the preferred method. Fourier transform and wavelet transform is a harmonic signal detection method most commonly used, most harmonic detection tools, power quality analyzers are based on these two theories, but the resolution of the Fourier transform of a single defect, wavelet wavelet transform over-reliance on the existence of small problems, leading them to the time-varying signal will appear larger deviations in the analysis, real-time detection ineffective. Neural network requires a lot of training samples, as well as non-standard sample selection problems, adaptive poor results.Secondly, this thesis introduces the basic theory of HHT, for EMD and Hilbert transform of the detail. By numerical example illustrates the HHT in nonlinear, non-stationary signals superiority that the HHT method aliasing modes exist, the end effect, false components and many other issues.Then, the problems for the HHT, this thesis presents an improved HHT method. For HHT modal aliasing problem, we use a digital filter to filter out the fundamental component, then the harmonic signal EMD decomposition, each IMF component, the final component to effective IMF Hilbert transform to obtain instantaneous frequency and amplitude thus the related parameter analysis, which can effectively solve the problem of modal aliasing; for end effects of HHT, analyzed the proposed continuation method sine function, advantages and disadvantages mirror extension and neural network extension method, we use mirror extension method, combined with the trend signal waveform transformation at both ends to effectively suppress the end effect of EMD problem. Then, the thesis by analyzing the simulation signal, the FFT method, wavelet packet method, improved HHT method are compared to verify the validity and strong resistance improved HHT method.Finally, the improved HHT method is applied to measure the oil rig system harmonic analysis. Current signal simulation analysis, efficient extraction of each harmonic current components, and comparative analysis with other methods, the results show a high accuracy of the proposed method has application value.