Infrared Image Analysis And Processing Of Power Equipmerts

Infrared thermal imaging technology is a non-contact measuring tern terature method. It is highly adaptable and has been widely used in online monitoring O power equipments.The infrared images gathered from the substations are analyzed and diagnosed manually based on individual engineers’ experience and knowledge.Recognition of different power equipments requires automatic proeessing due to a large number of gathered images. This paper presents a method wtich can automatically analyze and process infrared images of power equipments. Acco ding to the results of simulation, the proposed method has realized automatic them al fault judgment. It has also greatly improved the work efficiency and the recognition precision.Firstly.this paper uses median filter algorithm to pre-process the gathered images so as to filter border disturbance.This paper creatively uses K-means cl istering algorithm to achieve infrared image segmentation based on the temperature di ference between equipment and environment.The main aim is to extract the power equip ment in the infrared image.K-means clustering algorithm has addressed the shortcon ings of OTSU method that may bring error segmentation.The aim of image recognition is to automatically recognize the power equ pments in the infrared image and to meet the requirements of various image processing.Image recognition in this paper includes two segments:recognition of transfom er and recognition of three-phase power equipment.Recognition of transformer selects transformer as the research object.ln the light of its shape feature.this paper creatively uses x direction Canny edge detection algorithm to extract the image’s edge feature points.The paper determines the transformer’s edge through feature matching method based on maximum normalized correlation coefficient and K-means cl istering algorithm,finally uses least square method simulating line to identify the-specific location of transformer. Recognition of three-phase power equipmen uses morphological method to pre-process images,demarcates the connected region in the images,finally determines the region of each phase based on their area and bar/centric coordinate. This segment selects MOA and bushings as research objects.The aim of temperature recognition is to automatically obtain the m iximum temperature, minimum temperature and average temperature of the power equ pments according to color chart and to calculate the number and proportion of the pixels in each rempen ture range.As some pixels can’t exactly correspond to color chart.this paper uses pixel tlue Euclidean distance matching algorithm to assign the temperature value. Eventu:llythis paper has established tempcrature field of each power equipment.lost but not least.the air of fault recognition is to automatically determine the fault at ndition. fault type and its severity according to relev ant criteria.This paper has formed fault diagnostic proeesses of bushing.MOA transformer and has recognized the thermal fault of gathered intrared images.The simulations results have contirmed the effeetiv cness of the proposed method...