基于融合色差和神经网络的防震锤故障识别-《计算机技术与发展》

文章信息/Info

Title:: Fault Identification of Damper Defect Based on Fused Chromatic Aberration and Neural Network

Author(s):: TIAN Yi; WU Yi-qun; ZHANG Ye; HUANG Xin-bo^*; School of Electronics and Information,Xi’an Polytechnic University,Xi’an 710048,China

Keywords:: transmission line; damper; fused chromatic aberration segmentation; normalized cross-correlation function; Zernike moments; radial basis function neural network

摘要:: 防震锤是输电线路中抑制导线周期性振动及舞动的关键金具,提出了一种基于融合色差和径向基神经网络的防震锤缺损故障识别算法。该算法以无人机航拍输电线路防震锤图像为研究对象。首先,利用可保留原图颜色信息的单通道直方图均衡化增强图像;其次,提出一种基于图像颜色耦合性的融合色差算法,结合形态学处理,对增强结果进行图像分割,获取输电线路前景部分,并通过 Hough 变换对其中的输电导线进行标记,以确定可能存在防震锤的矩形区域;然后,以基于倒 T 型模板的归一化互相关系数(NCC)和Zernike矩分别被作为防震锤粗识别和精识别的依据,进行防震锤的定位;最后,以8维 Zernike? 矩特征作为径向基函数(RBF)神经网络的输入,实现输电线路防震锤缺损故障识别。实验结果表明,该方法对防震锤缺损故障的灵敏度较高、鲁棒性良好,识别准确率可达 91.67% ,为输电线路运维人员提供可靠的参考信息。

Abstract:: The damper is the key component to eliminate the periodic vibration and galloping of the transmission line. We propose an algorithm based on the fused chromatic aberration and radial basis function neural network to identify the fault of damper defect. The image of transmission line damper taken by UAV is regarded as research object. Firstly,single channel histogram equalization is used to enhance the image,which can retain the color information of the original image. Then,combined with the morphological processing and the fused chromatic aberration algorithm which based on the image color coupling,the enhancement results are segmented to obtain the foreground part. And Hough transform is used to mark the transmi-ssion line in the foreground to determine the rectangular area where the damper may exist. Next,the normalized cross-correlation coefficient(NCC) based on the inverted T-shaped template and Zernike moment are used as the basis of the coarse and accurate identification of the damper resp-ectively. Finally,the 8-D Zernike moment features is used as the input of radial basis function (RBF) neural network to realize the fault identification of damper defect. The experiment shows that the proposed method has high sensitivity and strong robustness to the fault of damper defect,and the recognition accuracy can reach 91.67%, which provides reliable reference information for the operation and maintenance personnel of transmission line.