基于Faster RCNN算法的输电线路防震锤识别研究

Identification of Transmission line Damper based on Faster RCNN Algorithm

针对输电线路人工检测金具成本大、效率低及后期分析样本时寻找缺陷困难等问题,提出了一种基于深度学习的输电线路金具识别方法。首先,通过无人机巡检获取一定数量的样本,通过将已标记处理的样本与相对应xml坐标文件进行联合扩充的方法,获得角度丰富化、数量多样化和部分噪声化的数据集;然后,采用Faster RCNN卷积神经网络算法对高重叠防震锤区块进行迭代合并,构建防震锤识别模型;最后,利用所提出的识别方法,对吉林省某电网公司无人机实际线路巡检得到的可见光影像数据集进行训练及测试。经过对模型的测试定位,实验结果表明,在复杂背景下所提出的基于Faster RCNN的输电线路智能识别算法,在该区域对于防震锤的检测精度达到94%以上。该识别方法为进一步对其他类型输电线路金具的检测提供了新的思路参考。

Aiming at the problems of high cost and low efficiency of manual inspection of transmission line fittings,as well as the difficulty in finding defects when analyzing samples in the later stage,a new identification method of transmission line fittings based on deep learning was proposed.First of all,a certain number of samples were obtained through UAV inspection,and a data set with Angle richness,number diversity and partial noise was obtained through joint expansion of the labeled samples and the corresponding XML coordinate files.Then,the Faster RCNN convolutional neural network algorithm is used to iteratively merge the high overlap anti-seismic hammer blocks to build the anti-seismic hammer identification model.Finally,the identification method proposed in this paper is used to train and test the visible light image data set obtained from the actual line inspection of UAV in a power grid company in Jilin Province.After testing and positioning the model,the experimental results show that the Faster RCNN based intelligent identification algorithm proposed in this paper achieves more than 94%detection accuracy for shockproof hammer in this area under a complex background.This identification method provides a new idea for further detection of other types of transmission line hardware.