The Zero Insulator Detection Based On The Infrared Imaging Technology

Infrared detection technology is one of the more advanced diagnostic techniques. It has the characteristics of no contact, high efficiency and accuracy. Infrared detection technology is widely used in military, transportation, medical, etc. With the infrared detection technology constantly developing, it has been widely applied in electric power system fault detection. At present, power sector has been using infrared thermal imager to inspect insulators in circuits. Researching the zero insulators detecting method based on the infrared imaging technology and intelligent recognition technology, can greatly reduce the workload of insulator inspection, decrease detecting safety accidents, lower the probability of insulator pollution flashover blackout accident, and be help to improve the safe level of power system operation. According to the difference between the law of fever in insulator under normal conditions and the law of heat under fault conditions, the insulator infrared image is obtained which can reflect temperature information. Image processing technology and pattern recognition technology are used to detect zero insulators.In order to extract the characteristics of insulator infrared image, and correctly identify the target, insulator string infrared image preprocessing is necessary. Insulator string infrared image pretreatment process includes image denoising, image enhancement, segmentation and angle correction. Image recognition process involves the extraction of single insulator disk, the extraction of feature vector, the design of classifier, the training of the classifier, and the validation of the identification model.Image pretreatment process consists of the following specific steps. First of all, the adaptive median denoising algorithm is used to remove the noise in the image. The adaptive average denoising can effectively remove noise and maintain the contrast between the target and background. Secondly, under different conditions, different insulator string infrared image has the different contrast. General enhancement algorithm will not be able to enhance the contrast of each image. The double threshold adaptive enhancement algorithm is used to strengthen image according to the distribution of grayscale in the image, eventually getting better effect and laying a good foundation for the subsequent image segmentation. Again, the improved 2d Otsu is used to segmenting image, obtaining the binary image of target. Finally, the edge detection algorithm is used to detect the target edge and the Hough transform is used for line detection, image angle calculating and image angle correcting.In the specific process of image identification, first of all, the single insulator disk is extracted by ellipse detection algorithm. Secondly, suitable temperature characteristic parameters are extracted and proved to be effective by the three factor analysis of variance significance testing. Again, a suitable classifier based on the requirements is designed. Once again, the classifier is trained. Finally, use the test data to identify the model and the accuracy of detection algorithm, proving the effectiveness of the algorithm.