The Research And Application Of High Voltage Equipment Infrared Recognition Technology Based On Scene Constraints

Normal operation of high voltage equipment in substations is a matter of people’s daily lives and the country’s economic development. Infrared detection technology can perform nondestructive testing of high voltage equipment which is widely used in major substation fault detection. Different types of high voltage equipment failure has different thermal principle and different external temperature,which means the thermal fault diagnosis parameters of the different types of high voltage equipment are not the same, therefore before thermal fault diagnosis of high-voltage equipment, the type of device needs to be determined to obtain the correct diagnostic parameters.This thesis focuses on the recognition of substation equipment and the corresponding thermal fault automatic diagnosis issues with infrared detection techniques, including recognizing visual equipment of online infrared camera constrained with the whole scene and handheld infrared camera constrained with the local scene. Therefore, we can achieve to monitor high voltage equipment intelligently and improve efficiency.Firstly, we studied the relevant application technology of infrared image including industrial infrared image detection; infrared small target detection and infrared image object recognition. Besides, we discussed deeply reasons and classification of thermal failure. We also studied methods of thermal fault diagnosis. We proposed an approach of automatic fault diagnosis for high-voltage equipment based on identification. Then we can get appropriate diagnostic parameters from the result, which can result in automatic thermal test.Secondly, we researched on methods of location and recognition for high-voltage equipment constrained with the whole scene. We constructed thermal panoramic of high voltage substation equipment, which including the infrared image acquisition, infrared image feature extraction, feature point-based infrared image registration, thermal fusion. In order to solve online identification of visual panorama of high voltage equipment, we proposed a method of online location and recognition for high-voltage equipment based on thermal panorama. Experimental results show that our method can recognize efficiently high-voltage equipment and gain location in panoramic vision based on the manual annotation panoramic vision. This has profound theoretical significance and practical significance.Next, We researched on classification and recognition for high voltage equipment constrained with local scene. We use pattern recognition method to identify the high voltage equipment constrained with local scene. In the view of the high-voltage substation equipment infrared thermal imaging characteristics, we proposed a method of morphological operations and the regional growth to realize segmentation. We studied feature descriptor of Histogram of Oriented Gradient(HOG) for infrared image classification and identification and common classifiers, including KNN and SVM. Experimental results show that the HOG features and SVM classifier can identify the classification of high-voltage equipment constrained with local scene effectively and efficiently.Finally, this paper introduced application background of the high-voltage substation equipment identification in online infrared monitoring system, and built intelligent online infrared monitoring system. Using software engineering design combined with the actual substation environment, we apply the system requirements analysis, architecture design and database design to develop a substation infrared intelligent online monitoring system. Our intelligent online monitoring system verify identification and diagnosis of high-voltage equipment constrained with the whole scene and the local scene.