基于无人机的输电网故障跳线联板识别

跳线联板是输电网中重要设备,其是否存在故障对输电网正常运行具有很大的影响。但由于现有的算法是对输电网中所有的故障用统一的方法进行识别,没有对各类故障输电设备进行专门的研究,导致故障跳线联板识别率低。为了高效识别红外视频图像中故障跳线联板,首先针对输电线的红外图像特征,采用改进的OTSU阈值分割图像对红外图像进行分割;其次,采用漫水法滤波分离各个连通域,运用形态学滤去小区域,填充大区域内的孔洞;最后,提取连通域的骨架,并从骨架图像中提取出USFPF特征,通过该特征识别的故障跳线联板。实验结果表明,识别故障跳线联板准确率为85.71%,漏检率为14.28%,误识别率为2.8%。该方法能够较好地识别故障跳线联板,具有较好的鲁棒性。

Recognition algorithm for fault jumper connection plate of transmission network based on UAV

Jumper connecting plate is an important part of electricity supply network. Jumper connecting plate's status has a significant impact on electricity supply network's proper operation. However, all the existing algorithms identify the fault of electricity supply network with a unified approach, so we have no specific approach on various types of transmission equipment failure and it will result in low recognition rate of jumper connecting plate. In order to recognize the fault jumper connecting plate efficiently, we use improved OTSU method for infrared image segmentation and use flood fill method to separate each segmentation area. Secondly, we delete small areas and fill holes through dilation and hole filling algorithm, then get connected area's skeleton by the skeleton extraction algorithm. Thirdly, we find Harris corner point in skeleton image, and calculate USFPF feature. Finally, through the slope value recognition we can discern jumper connecting plate's fault. As a result, the successful recognition rate for jumper connecting plate's fault is 85.71%, the miss rate is 14.28%, and the mistake rate is 2.8%. Experimental results show that the method has good results.