Arc Fault Risk Evaluation Based on Image Recognition

DC arc fault is easy to cause fire and endanger the normal operation of DC distribution systems and equipment. The arc fault protection approaches could be different according to the severity of the arc fault. In order to solve the above problems, an arc fault risk evaluation method based on the image recognition is proposed in this paper. Firstly, the images of DC arc fault are preprocessed to improve the clarity of the images. Then, the morphological structure characteristics and texture characteristics of the images are extracted, and the arc fault risk level is divided into three levels to achieve the evaluation of arc fault risk level. Finally, the real position of arc is determined by the binocular distance measurement. The binocular camera is calibrated with the checkerboard, after the intrinsic and external parameters of the camera are gained, the arc fault images are stereo corrected. The Block Matching (BM) algorithm is used to stereo match the corrected image to obtain the three-dimensional coordinates and depth of the arc. The experimental results show that the accuracy of arc fault risk evaluation is above 95%. The positioning error of arc fault is less than 2 mm, which has good accuracy.