Enhancing Arc Fault Detection in DC Series Circuits Using a Sample Entropy and High-Order Cumulants-Based System

The study of arc fault detection has gradually gained popularity in recent years. Singular value decomposition noise reduction is one of array signal processing techniques that can produce the desired ideal effect. In this study, the original data is first preprocessed using the singular value decomposition method, and the resulting smooth signal is then replaced for the sample entropy. We compare and examine the various properties of the present signal in the two time-domain states. The features of the array signal can be extracted using the high-order cumulant to characterize the complexity and nonlinearity of the signal. Finally, the collected feature quantities are used for arc defect detection using machine learning techniques. Verification reveals that the waveform fluctuation amplitude of the high-order cumulant of the current signal increases and the waveform becomes more chaotic when an arc fault occurs. The experimental results show that the DC series arc fault detection system based on sample entropy and high-order cumulant can effectively detect the arc Faults of DC series circuits in power systems, and has great practical application value.