Research On Detection Method For Arc Fault In Low Voltage System

According to statistics, the occurrence of electrical fires is on the rise in recent years. The electrical fires caused by arc fault occur frequently, which can do great harm to people’s lives and property. So the development and promotion of low voltage protective devices for arc fault are extremely urgent in the electrical industries. Arc fault detection is the soul of arc fault protection, because the reliability of detection methods can directly determine the effect of the protection. If the arc was taken as signal converter, the inputs are voltage and current signals, the outputs are the sound, light, heat, electromagnetic radiation and other signals. Therefore multiple detection methods can be gotten based on the input and output signals. However, the methods based on output signals are more suitable for fixed situations such as switch cabinet, and when detecting arc fault in wires, mostly detection methods concentrate on voltage and current signals. There are still challenges for series arc fault detection, because the fault current waveform in series arc fault is highly affected by the load impedance, especially in some electronic equipment, whose normal current waveform is similar to arc current waveform in resistive loads.The detection method of series arc fault was studied in this paper for the purpose of improving the reliability of detection. To achieve the goal, the arc fault experiment platform was firstly improved by being added an arc light sensor. And with this improvement, the arcing periods of the data can be marked by the light signal that received by the arc light sensor, which improve the reliability of arc fault detection method from a data perspective. Then an experimental scheme was designed referring to relevant standards. After the experiment data was collected, effects of load types and arc gap distance on the load current and the load-side voltage waveform characteristics were analyzed. And even more, the voltage and current changes in non-arcing cases such as the plug operation, starting process and intermittent failures were also be analyzed. The arc fault detection method was studied in this paper taking use of the random and unstable characteristics of current and load-side voltage. And after the signal of arc current was extracted, a wavelet threshold de-noising method was put forward and the cycle mean of the difference signal was selected as arc characteristic detection index. The analysis of the experiment data showed that the presented method had some value in the detection of arcing faults. In the final chapter, based on the difference of dispersion degree of load-side voltage’ cycle mean between normal state and arcing state, another detection method was put forward and analyzed using standard deviation of load-side voltage’ cycle mean as the arc characteristic detection index. At the last, the work that was done in this paper was summarized, and some suggestions were proposed for series arc fault detection researching.