【作者】 仝卫国；

【导师】 苑津莎； 李宝树；

【作者基本信息】 华北电力大学 ， 电工理论与新技术， 2011， 博士

【摘要】 输电线路担负着电力传输的重要职责,对它的定期巡检是实现输电线路状态检测的主要手段,对保证电网安全稳定运行具有重要意义。本文把航空摄影测量技术引入到日常的输电线路巡检工作中,提出了一种基于航拍图像的输电线路识别方法和状态检测方法,可提高输电线路巡检信息处理的实效性。本文研究了航拍图像的预处理方法,以及输电线和绝缘子的图像识别与状态检测方法,主要工作和研究成果如下:1.根据航拍图像的特点,研究了图像预处理方法,通过光学校正方法改善了由于光照条件变化引起的航拍图像亮度不均衡和对比度下降的缺陷;针对航拍图像噪声干扰大以及运动模糊严重的情况,提出了基于小波去噪的改进维纳滤波算法,通过引入像质评价函数,增强了维纳滤波效果,提高了图像复原的质量。2.航拍图像背景复杂且变化多样,通过对输电线特点的分析,提出了一种复杂自然背景下的输电线提取与识别算法,通过改进的Canny边缘检测算法初步完成图像的分割,然后通过设定长度阈值和宽度阈值,结合数学形态学方法和自适应Hough变换,实现了输电线的自动提取与识别。3.在输电线识别的基础上,提出了一种基于主动视觉的输电线弧垂测量新方法,将航拍图像采集看作是一个主动视觉过程,利用三维重构技术与GPS定位技术,构建了输电线的空间曲线模型,进而计算出弧垂的大小;分析了检测系统各部分对测量精度的影响,给出了相应的解决方案。实验结果验证了这种方法的准确性和有效性。4.通过对绝缘子基本特征的分析,给出了颜色特征、形状特征、不变矩特征以及小波系数特征对绝缘子的描述方法,提出了基于航拍图像多特征量信息融合的绝缘子识别方法,构建了基于RBF神经网络的信息融合系统,在此基础上实现了绝缘子的识别与状态检测。通过相关算法对绝缘子常见故障包括掉串、裂纹、表面污秽、覆冰和积雪等进行了故障诊断。实验结果表明,该方法正确率高、抗干扰能力强,可以满足实际应用的要求。更多还原

【Abstract】 Through power lines electric power goes where it is needed, it is laden with responsibility. Power line inspection at stated intervals is is an effective means of realizing status detection, and it is is of significance for the safe operation of power grid. In this paper, the aerial surveying technique is introduced to power line inspection, and a method for recognition and status detection of power line based on aerial images is proposed to impove the timeliness on the inspection information processing of power line. The preprocessing method of aerial image and the methods for recognition and status detection of transmission line and insulator are studied in this paper. The main work is done as follows.1. According to the actual situation of aerial image, the preprocessing method is studied. By using optical correction, it can be improved the brightness unbanlanced and contrast descending of aerial images because of the light conditions changing. Due to the serious noise jamming and motion blurring in aerial image, an improved Weiner filtering algorithm with wavelet de-noising is proposed based on the image quality evaluation function, and it can produce the better filtering effect and enhance the quality of image restoration.2. The background of aerial image is very complex, and the background objects will change their appearance under the different conditions. An algorithm is put forward to extract and recognize power lines from natural complex background: a developed Canny operator is used to realize preparatory image segmentation, and an adaptive Hough transform combined with mathematical morphology method is studied to extract and recognize the power lines by setting proper threshold values of length and width.3. On the basis of power lines extraction and recognition in aerial image, a novel method for transmission line sag measurement based on active vision is proposed. Considering the process of aerial image acquisition is an active vision process, by the application of 3D reconstruction and GPS technology, the space curve model of power line has been constructed, and then the sag of power line can be calculated with this model. After the influences on the measuring accuracy of all parts of measurement system being analyzed in detail, the methods to eliminate the influences are proposed and identified. The experimental results have demonstrated accuracy and effectiveness of this method.4. Analyzing of the essential features of insulator, the description methods of insulator are presented by using color feature, shape feature, invariant moment feature and wavelet coefficients feature. A new approach of insulator recognition and status detection from aerial image is put forward based on multiple features information fusion, and the information fusion system based on RBF neural network is accomplished. On the basis of information fusion system, the recognition and status detection of insulator has been implemented. By the corresponding algorithms, the common faults of insulator have been diagnosed, such as string breakage, pollution, crack, icing and snow covering. The experiment results show that the method has the advantages of high correct detecting probability, strong antinoise ability and a good application prospects.更多还原