Based on feature fusion and arc fault diagnosis method of support vector machine

In power systems, arc discharge phenomena arise when various power electronic loads are operating, potentially compromising the system’s stability and even posing a fire hazard. To address this challenge, this paper presents a novel DC arc fault detection algorithm that combines feature fusion with Support Vector Machines (SVM). The high-frequency characteristic signals of arc currents exhibit pronounced pulse patterns after detection and linear amplification. During an arc fault, the slope of the current waveform changes rapidly, either increasing or decreasing. By extracting both the high-frequency characteristic signal of the arc current and the slope-based feature of the current waveform, and performing feature fusion, an SVM-based diagnostic model is developed for accurate fault detection. Additionally, a parallel acceleration module is employed to expedite the diagnostic process. Experimental results demon strate that the proposed algorithm significantly enhances the accuracy of series arc fault identification under both single and combined load conditions, making it particularly well-suited for the performance requirements of smart power grids.