Analysis And Research On The Method To Detect Voltage Sags And Recognize Its Disturbance Sources

As a special kind of power quality problems, voltage sag gets close attention,because of its hazards, wide impact and frequent occurrence. Voltage sag for severalfrequency cycles will lead to the operation interruption of the sensitive equipmentsand result in huge economic losses. The detection of voltage sag characteristicparameters and the disturbance sources recognition are the basic work to inhibitvoltage sag and improve transient power quality problems. In this paper, voltage sagdetection method and disturbance sources recognition method are researched. Maincontents are as follows：Firstly, the paper systematically introduces the common voltage sag detectionmethods, including the methods based on RMS, peak voltage, the average ofsingle-phase voltage conversion, dq transform, short time Fourier transform, wavelettransform, Hilbert-Huang transform, mathematical morphology and fractal analysis.Theoretical analysis for all above methods is briefly executed, and parts of themethods are simulated.Secondly, in this paper, a real-time voltage sag detection method based onwavelet transform is proposed. The new method uses wavelet transform to detectvoltage sag starting and ending times and sag amplitude, and has innovationcompared with some methods that based on the wavelet transform just only to detectvoltage sag starting and ending times. The new method takes full advantage of thewavelet transform modulus maxima, namely, using the modulus maxima point toportray the starting and ending times of voltage sag, and using the size of themodulus maxima to calculate the sag amplitude. In addition, the new method gets abetter real-time through the sag signal preprocessing. The simulation results showthat this method has the feasibility and high detection accuracy in real-time.Finally, the mechanism of disturbance producing for three common voltage sagdisturbance sources(short-circuit fault, induction motor starting, transformeroperation)is analyzed. On Matlab/Simulink, the corresponding simulation modelsare build, and the voltage sag waveforms caused by the three disturbance sources aresimulated. The waveform characteristic indexes are given on the basis of summingup waveform characteristics of voltage sags resulting from three differentdisturbance sources. The voltage sag disturbance source decision tree is constructed, and the voltage sag disturbance sources are correctly recognized. The simulation testobtains a high correct recognition rate, and verifies the feasibility and effectivenessof the recognition method.