基于图像的换流站用避雷器仪表识别

A Recognition Method of Lightning Arrester Instrument in Converter Station Based on Image

针对电力系统中的换流站用避雷器仪表的特点,提出了一种基于图像的站用避雷器动作次数和泄露电流的识别方法。首先,采用基于透视变换的方法对采集到的避雷器泄漏电流仪表图像进行畸变矫正。其次,通过图像预处理进行图像增强获得相对比较清晰的表盘信息。再次,利用垂直投影法和基于运动项的改进BP神经网络完成避雷器动作次数数字式仪表的识别,采用减影法和改进的Hough(霍夫)变换完成换流站用避雷器泄漏电流指针式仪表的识别。最后,采用现场拍摄的型号为3EX5 050的西门子避雷器仪表对本文所提算法进行了验证。结果表明该算法能够快速、自动、准确的识别出避雷器仪表读数。该方法避免了人工读取效率低和精度差的问题,而且只需要很少的改动便可应用于换流站中其他仪表的智能识别。

According to the characteristics of the converter station lightning arrester instrument in the power system, putting forward a kind of recognition method of the converter station lightning arrester instrument based on image. Firstly, doing a distortion correction on collected arrester leakage current meter image based on the method of perspective transform. Secondly, according to Image preprocessing for image enhancement to get relatively clear dial information. Again,using vertical projection method and the improved BP neural network based on sport to complete the identification of lightning arrester action times of digital instrument, then Subtraction method and improved Hough transform are used to finish the identification of converter station lightning arrester leakage current of pointer type instrument. Finally, the lightning arrester instrument of the 3EX5 050 were conducted to validate the algorithms proposed in this paper. The results shows that the algorithm is able to identify lightning arrester meter readings fast, automatically, accurately. And this method can avoid the low artificial reading efficiency and the problem of poor accuracy, but only can be applied to intelligent recognition of other pointer type instruments in converter station with very few changes.