Zhizhou Bao / People's Electrical Appliances Group Co., Ltd.
Yao Wang / Hebei University of Technology
The occurrence of low-voltage AC series arc faults will cause the temperature at the fault to rise rapidly, which can easily lead to electrical fires and cause serious losses to individuals and society. However, the detection accuracy of traditional arc fault methods is insufficient and cannot effectively curb the occurrence of arc faults. Artificial intelligence-based technology provides high-precision detection solutions, but the AI model itself is a "black box". Once a misjudgment occurs, the root cause of the model error cannot be fundamentally identified, and further improvements in model accuracy are limited. In order to solve the above problems, this paper proposes a new method for AC arc fault detection based on attention mechanism. The introduction of the attention mechanism effectively handles the weight between the input arc data and the model output, thereby improving the accuracy of model detection. Experimental results show that the model proposed in this article achieved a detection accuracy of 99.69%, proving the efficiency of this method.