【作者】 王小朋；

【导师】 孙才新；

【作者基本信息】 重庆大学 ， 电气工程， 2009， 硕士

【摘要】 在输变电系统中,输电线路的覆冰现象十分普遍。覆冰可以引起导线舞动、断线、倒塔及绝缘子闪络等重大事故,造成巨大的经济损失和一定的社会影响。因此,实时掌握输电线路的覆冰情况特别是覆冰厚度以便电力部门采取相应的措施避免输电线路覆冰引发重大事故成为人们目前研究的热点。针对现有输电线路覆冰厚度检测方法的不足,提出了一种新的输电线路覆冰厚度在线监测方法,即通过高压铁塔上的工业摄像机在线采集输电线路覆冰前后的图象,然后通过GPRS网络和Internet网络传送到电力系统控制中心,控制中心计算机系统对图像进行处理,提取其边界轮廓,最后通过将导线和绝缘子覆冰状态与非覆冰状态时的边界进行比较并通过一定的标定计算方法得出其当时的覆冰厚度。本文完成了在线监测系统的总体方案设计,并根据系统的特点设计了工业摄像机、GPRS图像传输模块以及监测端电源等硬件设备的方案和类型。本文通过试验及结果分析发现,将彩色图像转换成黑白图像进行处理完全满足本课题的需要,同时通过多帧相邻图像求平均值、调整灰度分布以及中值滤波等可以明显增加图像对比度、降低图像噪声干扰;应用小波变换和浮动阈值方法可以准确地提取边界较平滑的导线和绝缘子覆冰前后的边界轮廓,且依据该边界轮廓计算的导线和绝缘子左右部分的覆冰厚度的误差分别是0.1mm、0.6mm和0.4mm,均小于本文的单个像元大小(0.67mm),覆冰厚度计算精度较高;应用最优阈值和数学形态学方法能准确提取导线覆冰前后的边界轮廓,且该方法特别适用于提取导线覆冰边界起伏较大的情况,依据此边界计算的覆冰厚度能更好的接近导线的真实覆冰厚度,该方法同时解决了导线冰层透明时边界轮廓提取的难题。试验同时发现,本文提出的方法对导线的覆冰厚度计算精度要比绝缘子略高,且导线和绝缘子静止状态下的覆冰厚度计算误差一般不会超过图像单个像元的大小,因此提高输电线路图像的分辨率,降低像元的大小可以显著减小覆冰厚度的计算误差。本文根据摄像机成像公式,提出采用振动覆冰导线在距离摄像机远近不同位置时计算的覆冰厚度中间值作为振动导线的真实覆冰厚度,该方法在实验室验证条件下的误差是1.5mm,在覆冰厚度不发生明显变化的前提下尽量多采集振动覆冰导线的图像可以降低覆冰厚度的计算误差。更多还原

【Abstract】 In the power transmission system, the icing phenomenon of transmission line is very common. It can cause conductor galloping, tower falling down and insulator flashover, which may cause power failure, affect people’s normal production and life and result in great economic loss and a certain degree of social influence. Therefore, to master the icing situation of transmission line, especially the icing thickness real-time so that the electricity sector can take appropriate measures to prevent transmission line from a major accident is becoming a research focus.A new method of measuring the icing-thickness of transmission line on-line is proposed in this paper for the deficiency of existing methods. In this method, the pictures of transmission line which are photoed by the camera on the iron tower are transported to the control centre of power system through GPRS and Internet network firstly, and then, the pictures are processed there to extract the edges of conductor and insulator. The icing thickness can be gained by comparing the iced edge and the uniced edge. The type and configuration of industrial camera, GPRS module and power device are designed according to the characteristics and requirements of the proposed monitoring system in this paper.Through analyzing the test and its result, the conclusion can be obtained as follow: converting the original RGB images to gray images can satisfy the needs of the subject in this paper. Calculating the average value of some adjacent images, adjusting the gray level distribution and using median filtering can increase the contrast and reduce the influence of noises in the image; the method based on wavelet transform and floating threshold can extract the edges of conductor with smooth icing boundary and insulator accurately. The experimental results show that the icing thickness calculation errors of conductor, the left part and the right part of insulator which are based on the edges extracted through wavelet transform and floating threshold are respectively 0.1mm, 0.6mm and 0.4mm which are all less than the size of a single pixel (0.67mm in this paper) and the calculation accuracy is satisfactory; the method based on the optimal threshold and mathematical morphology can extract the edges of iced conductor and uniced conductor accurately and is especially suitable for edge extraction of conductor with big ups and downs of icing boundary. The problem of the transparent icing edge extraction on conductor is also solved successfully in the method; the experimental results also show that the errors of the icing thickness calculation values which are gained through the proposed icing thickness calculation methods in this paper are normally less the size of a single pixel. So improving the resolution of images and reducing the size of a single pixel can significantly reduce the icing thickness calculation error.The method that using the median value among the icing thickness calculation values through the vibrating conductor images in different location as the true icing thickness of the vibrating conductor based on the camera imaging fomula is proposed. The icing thickness calculation error of vibrating conductor based on the method in this paper is 1.5mm and the error can be reduced by acquisting more icing images of vibrating conductor to calculate the median value of pixel numbers corresponding to icing conductor diameter in different places on condition that the icing thickness doesn’t change significantly.更多还原