Study On Detection Method Of Series DC Arc Fault

With the rapid development of power electronics technology,the DC power supply system has been further promoted and applied.However,at the same time,problems such as aging of the insulation of power supply line,loose joints,or poor contact resulting from broken connection,can cause series DC arc faults and endanger the power supply system.The fault arcs existing in DC systems such as photovoltaic power generation system and energy storage system are different from AC arc faults and do not have the zero-break characteristic of AC arc faults.The current value of series arc faults is less than the loop current value under normal operating conditions and the current value of parallel arc faults.Protection devices such as circuit breakers cannot effectively test the series arc faults.Therefore,accurate detection and location of series DC arc faults play an important role in ensuring the safe operation of the power supply system.The paper introduces,experimentally analyzes and compares three detection methods of series DC arc faults.The first is the series DC arc fault detection method based on arc current characteristics.Sampling the arc current through Hall current sensor and current transformer respectively.The time domain waveforms and frequency domain waveforms were analyzed to extract time domain features and frequency domain features to achieve multi-feature fusion.Sampling different types of data under different experimental conditions and dividing these data by fuzzy C-means clustering.The results of sensitivity verification indicate that this detection method has low sensitivity under complex working conditions,especially when the inverter is running.An arc faults detection method based on the characteristics of the fault arc conduction interference signal was proposed to improve detection accuracy.In order to reduce the misjudgment rates,a multi-segment frequency selection network was used to obtain frequency features including multi-segment frequency domain amplitude and energy spectral density integral value.The sensitivity experiment under different working conditions and the attenuation experiment of high-frequency characteristics with the transmission distance were carried out separately.The recognition results indicate that this detection method has significantly improved sensitivity compared with the common Hall current detection method and current transformer detection method.The shortcomings of the series DC arc faults detection method based on arc current characteristics and the coupling problem existing in multi-loop circuits when detection method based on characteristics of conduction interference of series DC arc faults were analyzed.This means that the lack of location ability of detection method based on conduction interference features can result in erroneous judgements in multi-loop circuits.Therefore,a detection method combining the high-frequency characteristics of the arc branch current and the time-frequency characteristics of the conduction interference of arc faults was proposed to realize the location of arc branch in multi-loop circuits,which called arc branch combination identification method.A wide variety of experiment platforms involving different resistance loads,the high-pressure sodium lamp,switching power supplies,the halogen lamp,different photovoltaic output powers,different electrode materials and the inverter operation were built to compare the sensitivity of different arc fault detection methods introduced.Misjudgment analysis under different experimental conditions was displayed to verify the effectiveness of the proposed detection method.