Study On Transformer Substation Safety Incident Automatic Identification Technology Based On Video Image

The transformer substation is defined as important hub of the electricity power production. The video supervisory control systems have realized intuitive monitoring of environmental conditions, equipment operation, repair and maintenance of various types of production safety situation through the method of artificial polling images and visual judgment. According to the missed checking event, hysteresis problems of current manual inspection in transformer substation, the paper uses the video collected from the transformer substation to research and develops a security incident automatic identification alarming system based on the digital-image processing technology in MATLAB7.8, the system can realize monitoring transformer substation’s safety events with all-weather, intelligent and automatic detection, reducing the workload of monitoring personnel and reaching monitoring, alarming real time of all events. The system enhances transformer substation’s safe operation of the emergency support capabilities, and has important scientific significance and value of projects. The specific research works are shown as follows:1. Based on linear features and color features, the paper researches two cordon detection technology of transformer substation. The method of combination edge detection and mathematical morphology realizes detection of line based on Hough transform principle. The method of combination median filtering and morphological operations realize to extract of the color information based on image binarization.2. Static and dynamic smoke detection technology. The paper extracts the smoke color, brightness, saturation characteristics etc., and realizes static smoke detection through combining the analysis of morphological operations and connectivity; through extracting and processing exact frame images determining the appropriate threshold based on background subtraction of moving target, researcher realizes dynamic smoke detection effectively, and reaches the effect of expected detection.3. Studying on character detection and personality tracking. The paper realizes personality detection based on background subtraction, designs and proposes particle filter character tracking algorithm based on color features. Improving the color features selected methods and heavy particles the right calculation method can make tracking trajectory diagram show tracking the position accurately, the result of time tracking performance is perfect. The combination of tracking algorithm and figure detection realizes characters automatic detection and alarm, the results are satisfactory.4. The paper researches and develops an automatic detection and identification of the substation security event video and alarm system in MATLAB7.8. Through verifying of existed video that are collected from the transformer substation, the system can be accomplished the formation of video conversion effectively, image preprocessing, cordon detection, smoke detection, character detection and tracking, and automatic alarm of event etc.