Research On The Localization And State Detection Of Insulators In Aerial Image

Insulator is an indispensable element on transmission lines with the dual function of wire support and electrical insulation. Frequent faults of insulator, such as filth, crack, and breakage affect the normal operation of transmission lines, shorten the operating life of lines, or even worse, may lead to large-scale blackout and huge losses. Therefore, it is necessary to real-time monitor the status of insulator. And the localization of insulator in aerial image is the prerequisite of status detection and fault diagnosis of insulator. Therefore, this research includes two main problems: the localization and the status detection of insulators in complex aerial image.This paper studies the localization method based on feature points, and proposes the localization method of multiple insulators based on SURF and IFS; to solve the problem of high localization uncertainty caused by experience value recognition method, and realize the localization of multiple insulators with multiple main directions, wide-angle deflection, this paper proposes the localization method of multiple insulators based on main direction detection and equidistant model of shape feature points; for the complex computation caused by prior main direction method,this paper proposes the localization method of insulators based on collinear and isometric constraints. This method uses collinear and isometric constraints of curvature scale space points, and realizes the localization of insulators with any main direction in complex aerial image.After positioning, divide insulator into multiple umbrella plate as test image.This paper builds the aerial image database composed by normal, defect, crack and noise insulator umbrella plate images, trains convolution neural network, and classifies the 4 types of insulator umbrella plate images. This method can classify multiple insulator images with different superficial failures simultaneously, and has high classification accuracy.The automatic localization of insulator can lay a foundation for the real-time detection and fault diagnosis of the running status of insulator. The status detection method of insulator can realize the intelligent fault diagnosis to avoid heavy losses.Research on the localization and status detection of insulator in aerial image realizes the automatic localization and fault diagnosis, and can guarantee the safe and reliable operation of power grid.