Design And Implementation On The Name Plate Recognization System In Substation

The substation is an important part of the power system, which is not only undernormal mechanical and electric load, but also endure pollution, lightning, strong winds,bird damage etc. It’s necessary to inspect important substations and lines regularly orirregularly to prevent equipment failure and accident. In recent years, with theadvancement of technology and reform in the power system, the human inspection isgradually being replaced by the robot inspection, which is also the future trend of smartgrid.Against the above background, Chongqing Electric Power Corporation (CEPC)proposes the substation inspection robot research project. CEPC has10500kvsubstations and65220kV substations which are crucial to Chongqing grid. Theinspection to all the facilities in the substations is an very burdensome task. Theinspection robot can automatically finish a global path planning according to theinspection tasks setting, and record the status of electrical facilities and abnormal alarmby sensors. Recognizing equipment name board is vital for inspection robot todistinguish electrical facilities in substations. The aim of the paper is to realize anelectrical facility name board recognition system for inspection robot.Name board recognition system has3parts: name board locating, name boardimage transmission, name board character recognition. Based on the analyzing to themobile image acquisition technology, this paper puts forward a name board locatingalgorithm which implemented based on OpenCV’s C library with the help of Javalocalized interface. The algorithm has higher identification accuracy and real-timeefficiency. On the image data transmission, a method which use java media framework(JMF) to capture the image frame for compression and transmission from the videostream is proposed. The method has many advantages, such as cross platform, easy todeployment, and low cost. In the research on name board characters segmentation andrecognition, a three steps method is proposed. Firstly based on the analysis to slantcorrection algorithm., color space conversion is used to complete gray conversion andhistogram equalization, and P-tile method is used to select an optimal threshold value for binarization; secondly the outline of characters are determined by projection method;lastly template matching algorithm is used to character recognition. The recognitionaccuracy is enhanced by improving the templates.The achievements of this research have been used in the substation inspection ofChongqing Electric Power Corporation, and its effects meets the needs of practicalapplication.