Power distribution facility switch detection and recognition algorithm based on deep learning

Abstract

The invention discloses a power distribution facility switch detection and recognition algorithm based on deep learning, and the algorithm comprises the following steps: S1, collecting a photo of a to-be-recognized knob switch, and marking the position of the knob switch in the photo; S2, building a deep learning neural network by adopting a preset method, performing feature extraction on the marked photos, and storing a trained model; and S3, using the trained deep learning neural network model to recognize a to-be-recognized picture, and detecting the positions of the knob switch and the rotation center and the pointing point of the knob switch. The invention has the advantages that the position of the knob switch can be corrected, the number of times of photo shooting is reduced, the recognition efficiency is improved, even the knob switch on the edge of a picture can be recognized, the requirement for distortion of a camera is not high, the cost for purchasing the camera is reduced, in addition, the invention can be used for various occasions where the knob switch state needs to be detected. The state of the knob switch can be detected in real time, and alarm information is sent out when the state is abnormal.

Description

technical field [0001] The invention relates to the technical field of recognition algorithms, in particular, to an algorithm for detecting and recognizing switches of power distribution facilities based on deep learning. Background technique [0002] Electrical equipment is the basic component of the power system and the basis for ensuring the reliability of power supply. In order to facilitate the control of electrical equipment, it is often necessary to set a number of control buttons or knob switches on the control panel of the electrical equipment. At the same time, in order to ensure the smooth operation of the electrical equipment, it is necessary to detect the control buttons or knob switches on the control panel in real time, so as to judge whether the electrical equipment is operating normally. [0003] At present, the traditional rotary switch detection methods generally have the following methods: 1. Visual inspection method. This identification method is only su...

AI Technical Summary

Problems solved by technology

2. The rotary switch recognition method based on the traditional image algorithm. This recognition method has a slow recognition speed and cannot accurately identify the direction of the rotary switch. It may also recognize other objects on the device as a rotary switch. The image quality requirements are high, the algorithm robustness is low, and there are many false detections and missed detections

Images