Research And Application Of Chongqing Lanjiatuo Substation Video Surveillance Technology

With the application of automation technology in grid system, substation surveillance mode is gradually tended to be less people or unmanned. Among them, the unmanned substation is becoming an important part in the development of grid technology. In the advancement of unmanned mode, the role of "remote-viewing" is getting more and more attentions. Remote-viewing serves as the eyes of the unattended substation, and it is an integral part of substation information collection system. However, with the deepening of unmanned substation, the shortcoming of traditional remote-viewing function reveals constantly.Current substation video remote monitoring system is still mainly worked as traditional video surveillance, only provides simple functions such as video capture, storage and playback, which lacks the ability to effectively analyze information. Besides, the monitor of substation equipment is still relying on artificial observation and patrol, which did not take full advantage of the potential of intelligent video surveillance technology. On the side of information storage, current remote-viewing system not only wastes the hard disk space, but also goes against to the retrieval of the video.Intelligent video surveillance technology involves digital image processing, machine learning, pattern recognition, signal processing, statistical learning, and many other interdisciplinary areas. It possesses important research value and broad application prospect. Integrating the intelligent video surveillance technology with the existing substation remote viewing system, the functions of remote- viewing system can be improved and enhanced even without need to make big changes to hardware equipment. This thesis is work with 110 Kv substation remote monitoring system in Chongqing Lanjiatuo and algorithms based on SVM and ViBe will be applied. The main contents consist of two parts:Firstly, online-monitor method of the equipment with obvious color features is proposed. The working states of some equipment can only be judged by their color, owe to no other significant changes in texture, contour, shape. By extracting the color feature information, color classifier could be trained and equipment can be monitoring automatically. Whenever exception is occurs, worker will be reminded to replace these equipments. These operations not only reduce human consumption, but also reduce the burden on staff.Secondly, multi-pathways video surveillance for moving target detection method under outdoor environment is proposed. So many cameras are equipped in substation, and all need continuous monitoring for long time. In order to get a real-time algorithm, we need to reduce the complexity of the algorithm as much as possible. Besides, under the outdoor environment, light, weather, occlusion, shadows and other circumstances are changing frequently. Moving target detection algorithm is difficult to achieve good tradeoff between accuracy and efficiency. After the improvements of original Vi Be,, we achieving better performance of moving target monitoring,which still with low complexity.