Localization And Analysis Of Transient Power Quality Disturbance Based On Hilbert-Huang Transform And Wavelet Packet Denoising

Accurate and fast localization of the disturbance begin-end time by the transient power quality detection method is the key to determine the quality of a certain method. The purpose of the thesis is to realize the localization and analysis of the begin-end time of five kinds of typical disturbance (including voltage interruption, voltage swell, voltage sag, voltage flicker and transient oscillation). First, denoises the noisy disturbance signal by wavelet packet denoising. Then compares between wavelet transform detection method and Hilbert-Huang transform detection method, and selects one method with stable and accurate detection performance. All the modeling and simulation is realized by MATLAB.The main content and research result is as follows:(1) Wavelet denoising method and wavelet packet denoising method are used to denoise the noisy signal respectively. Compares the simulation result of different decomposition layers:3,5 and 7 layers. As the number of decomposition layer increases, the denoising effect shows a bad-good-bad trend, and the wavelet packet denoising method is better than the wavelet denoising method. Both methods achieve best denoising effect at 5 decomposition layers.(2) In order to further determine the best denoising function of the wavelet packet denoising method, we choose eight functions of db1, db3, db5, db7, sym1, sym3, sym5, sym8 and compare among their denoising effect. We select four functions of db3, db5, sym5, sym8 with relatively better effect and further compare among these four functions. Combining (1) and (2), we conclude that function sym5 has the best denoising effect at 5 decomposition layers.(3) Two methods are used for the transient power quality disturbance localization. The first one is wavelet transform detection method that conducts detection based on the principle of wavelet algorithm. The second one is Hilbert-Huang transform detection method. Use these two methods respectively to detect the begin-end time of disturbance, and compare between the detection results. As for the noisy signal detection, the wavelet algorithm fails to detect accurately three kinds of disturbance:voltage flicker, transient oscillation and voltage sag. The algorithm also fails to detect the four disturbance points set by the voltage flicker, which decreases the accuracy rate of the transient power quality disturbance detection.(4) Hilbert-Huang Transform (short for HHT) detection algorithm is adopted to locate and analyze the typical transient power quality disturbance signals such as voltage swell, voltage sag, voltage interruption and voltage flicker. The experiment result proves that the algorithm is able to achieve accurate disturbance location result and validates the feasibility of the algorithm. The disturbance detection accuracy is further improved by combing the wavelet packet denoising algorithm. It also provides more accurate and reliable support for the power system disturbance control.