The distribution network line is low in height and densely laid.Due to the interference of external factors,it is easy to be damaged by external forces,therefore,an improved Faster-RCNN distribution network line anti-outburst detection method is proposed.Firstly,image enhancement and annotation are carried out for the limitations of the collected dataset,and a high-quality image data set for anti-outburst of distribution network lines is established.Secondly,in order to improve the feature extraction ability and feature learning ability of Faster-RCNN,the deep residual network ResNet101 is used to replace VGG16,which is integrated into the CBAM attention mechanism module,and the feature pyramid is introduced for structural improvement.Then,in order to improve the learning effect,the difficulty sample balance loss function optimization is used to improve the parameters.Finally,the image is verified by collecting images in a certain area of Northwest China.The results show that the proposed model has strong robustness,good generalization and certain advantages.
关键词
配网线路/防外破检测/改进Faster-RCNN/结构改进/参数改进
Key words
distribution network line/anti-outburst detection/improved Faster-RCNN/structure improvement/parameter improvement
维普
万方数据