Detection And Recognition Study On Power Quality Disturbances

In recent years, with the processes of electricity market and environmentalrequirements, safe operation of the power grid has attracted more importance. The stabilityof power system has been prominent. There insist a series of power quality problems haveto threat the power quality, which include voltage sag, swell, interruption, impulse,oscillation transient, harmonics, voltage flicker and so on. Meanwhile, the reduction of thepower quality has brought inconvenience to people’s life. Hence, it is very significative toimprove power quality. And Power quality detection and recognition is the premise toimprove power quality. Thus, the dissertation researches on three aspects, about thetransient disturbances detection, single disturbance recognition and multiple disturbancesdetection and recognition.About the disturbances detection, an improved method of selecting threshold is usedto determine the modulus maxima. The single disturbance and multiple disturbances aredetected respectively. The simulation results show that the proposed method has goodresistance to noise; and it can accurately determine the start time and end time (includingzero-crossing disturbance) of the single disturbance. At same time, it has good adaptabilityof the sag and harmonic, interruption and oscillation.About single disturbance recognition, the dissertation presents a novel algorithmwhich combined the dynamic time warping (DTW) and wavelet boundary descriptors(WBD). Firstly, we got the wavelet boundary descriptors to extract profile feature of thosedisturbances; secondly, according to certain rules, DTW is used to match testing and sevenkinds of reference disturbances; finally, it calculated the distances matrix to obtain theshortest distance in two time-series model and determine the similarity of contour. Thesimulation results prove that this method is not affected by amplitude and duration. Also itgets a good recognition result. At the same time, it has good and high recognition rate ofthe voltage flicker and impulse.About multiple disturbances recognition, the dissertation presents a novel methodbased on decoupled multiple synchronous reference frame PLL (DMSRF-PLL) andwavelet. In the first place, wavelet filter is used in disturbances filtering. In the next place, decoupled multiple synchronous reference frames PLL is used to sample the fundamentalcomponent and frequency of the signal. This method can be used to determine whether thesignal contains harmonics, sag, swell or interruption. Finally, it detects the number ofextreme value point to identify whether it is composed of impulse or oscillation transientdisturbance. The method respectively identified lots of single disturbance and multipledisturbances and show that this method has excellent capacity of discernment for multipledisturbances. In the meantime, the method can realize a grid voltage synchronousdetection disturbance signal. Therefore, it has certain practical value in power qualitydetection and recognition.