Shared convolutional neural network-based insulator detection method and apparatus

Abstract

The invention provides a shared convolutional neural network-based insulator detection method and apparatus. The method comprises the steps of shooting a power transmission line image of a substation by utilizing a patrol robot; and obtaining an optimal position of an insulator in the power transmission line image by utilizing a shared convolutional neural network obtained by training an RPN and a Fast R-CNN. According to the method and the apparatus, the insulator in the power transmission line image is detected by utilizing the shared convolutional neural network obtained by training the RPN and the Fast R-CNN of a shared partial convolutional layer and a pooling layer; and compared with the prior art, the calculation complexity can be lowered, real-time detection of the insulator under a complex background is achieved, and accurate identification and locating of the insulator in a robot patrol image are realized.

Description

technical field [0001] The present invention relates to the field of target detection of key power equipment of power equipment, and more specifically, to an insulator detection method and device based on a shared convolutional neural network. Background technique [0002] In recent years, ensuring the reliability and operation of transmission lines has become an important part of building smart grids. The safe operation of substation equipment is the prerequisite for ensuring the stability and safety of the power system. Insulators are indispensable insulating components for power transmission lines, and their operating conditions directly affect the reliability and safety of the power grid. At the same time, the insulator plays the role of electrical insulation and support in the transmission line; and its surface pollution, cracks, damage and other problems seriously threaten the safe operation of the transmission line. According to statistics, the accidents with the hi...

AI Technical Summary

Problems solved by technology

[0004] Due to the complex background of the insulator pictures taken by inspection robots and drones, most of the traditional methods of detecting insulators require manual inspection one by one, which not only consumes a lot of manpower and material resources, but also easily causes casualties, and is prone to missed and misjudged phenomena.

[0005] At present, although some automatic identification methods for insulators have appeared, these methods are prone to misidentification results because of the complex background of substation images and the existence of other electrical equipment similar in shape to insulators in the images, such as current transformers and lightning arresters; Or when the amount of data increases, it will greatly increase the computational complexity and calculation time, which cannot meet the requirements of real-time detection

Images