Research On Arc Fault Current Characteristics Of Arc Fault Circuit Interrupters

The occurrence frequency of electrical fire is higher in China and the loss caused by the electrical fire assumes a leading place among all the fires, while the arc fault is one important reason to cause an electrical fire. In order to prevent the arc fault form causing electrical fire effectively, the arc fault circuit interrupter emerges as the times require. Arc fault circuit interrupter (AFCI) can find out arc fault, and automatically cut off power supply to avoid causing more serious accidents. Its key function is accurate discrimination of the arc fault. The purpose of this thesis is to investigate the basic differences between arc fault current and normal current. Different mathematical methods are used to analyze the characteristics of them, and identify whether an arc fault appears according to the analytic results.This thesis analyzes the relationship between arc fault and electrical fire, and compares the difference between AFCI and other circuit breakers. An experimental platform is built for investigation on the fault arc characteristics. The loads contain resistance, dimming lights, air conditioner and computer. Under these loads, the experimental data are measured for series arc fault and normal currents.According to the measured data, a parameter analysis is made to compare their differences in valid value, mean, peak to peak, flat shoulder percentage and rate of current1increase. Fourier transform is used to inspect the frequency spectrum characteristics. A comparison is made for the difference of odd and even harmonic contents respectively under in fault and normal operating conditions. Furthermore, wavelet transform is also used to analyze the measurement data. Based on the error values obtained from the decomposition-reconstruction, a reasonable choice is made for the wavelet basis function, by which noise processing, failure point judgment and energy feature vector extraction of wavelet transform are performed respectively. The feature vector is used as the input sample of wavelet neural network. Finally, a neural network method is proposed for identifying the arc fault, which is based on the results from integrated index analysis, Fourier transform and wavelet transform. The related flow chart is also given. The neural network training and testing are carried out for the data, the obtained results show that the wavelet neural network has a higher identifying rate, and the wavelet neural network can provide a new idea for the arc fault detection.