Algorithm for Digital Instrument Recognition of Substation Inspection Robot

Abstract

The invention discloses a digital instrument recognition algorithm of a substation inspection robot, which includes: the substation inspection robot collects equipment images, and preprocesses the images; automatically locates the digital area; corrects the inclination of the digital area; Segmentation operation, digital positioning is performed according to the contour obtained by segmentation, and a single digital image is obtained; digital images of various forms on the substation site are collected, a digital recognition training sample set is established, and a classifier is integrated; a single digital is recognized using a trained integrated classifier, The final recognition result is obtained by sorting the numbers according to their coordinates in the image. Beneficial effects of the present invention: even when the surface of the on-site meter is dirty and has a certain inclination angle, a good recognition effect can still be obtained, and digital meters with certain incompleteness can also be correctly recognized, and can be adapted to various substations. different types of numbers.

Description

technical field [0001] The invention relates to the technical field of substation digital instrument identification, in particular to a digital instrument identification algorithm for a substation inspection robot. Background technique [0002] Many substations in the power system are equipped with digital instruments, which are widely used because of their simple structure, intuitive, easy operation, low cost and low failure rate. A large amount of data needs to be recorded every day for such meters in the substation. At present, domestic instrument data is mainly recorded by human eyes, which has low precision, poor reliability, low efficiency and high labor intensity. With the pilot construction of smart substations and the development and popularization of inspection robots, the recording of instrument data has been improved towards automatic operation, but operators still need to participate in the verification and confirmation of instrument status images collected by ...

AI Technical Summary

Problems solved by technology

[0004] 1. At present, most of the digital recognition systems operate in an ideal lighting environment. Once they break away from this premise, the recognition effect is difficult to guarantee

[0005] 2. The feature matching and endpoint detection methods used in many schemes have high requirements for the segmentation of digital areas, and the robustness and scalability of such methods are greatly limited

[0006] 3. Most of the existing schemes deal with numbers in a single form, and cannot effectively deal with numbers in various forms in substations

Images