DC series arc faults are one of the main causes of fire hazards in photovoltaic power systems. The common method of the traditional dc series arc fault detection uses wideband current sensors to obtain the arc current signal, extract arc characteristic frequency components, and make intelligent judgments based on numerous samples. This kind of detection method requires high-performance hardware and has a high cost. This article takes the common-mode conductive voltage (CMCV) signal, which is produced by arc faults, as the object and confirms the feature frequency band of the arc CMCV signal based on the spectral analysis. To obtain the feature frequency component and perform frequency reduction at the hardware level, multiple frequency selection and detection circuits were designed, and the feature quantities of the time and frequency domains were extracted from the output of the circuits. Subsequently, a two-dimensional arc fault feature plane was constructed, and an arc/normal similarity criterion and arc fault judgment algorithm were proposed. Finally, this article discusses or experimentally verifies the effectiveness, reliability, and anti-interference ability of the proposed method. The proposed method is low cost, simple, reliable, and effective and provides a novel idea and approach for dc series arc fault detection.
Read full abstract