Method for modeling image characteristic data of power equipment

Abstract

The invention discloses a method for modeling image characteristic data of power equipment. The method comprises the following step: after the pictures of the power equipment collected in the regularpatrol process of the intelligent equipment of AR wearing equipment or a patrol robot are recognized, carrying out definition specification of the image characteristic data of the power equipment withuniform appearance by adopting a D language to form structured table data, wherein the power equipment comprises primary field equipment, secondary field equipment and metering field equipment. According to the method, the collected image characteristics of the power equipment are subjected to unified and standardized data modeling, so that on the basis for forming large data of an image libraryof the power equipment, the fault statistics of the power equipment are facilitated, the intelligent identification accuracy of the power equipment is improved, the operation and maintenance effects of the power equipment are greatly improved, the safe and stable operation of a power system is further guaranteed, the structured data is converted, the image data is greatly compressed, the image data is conveniently stored in a database, and mining analysis of large data in later period is facilitated.

Description

technical field [0001] The invention belongs to the technical field of knowledge databases for operation and maintenance of power metering equipment, and in particular relates to a modeling method for image feature data of power equipment. Background technique [0002] With the continuous expansion of power grid scale and equipment, intelligent inspection equipment based on artificial intelligence technologies such as speech recognition, image deep learning and image recognition technology is becoming more and more popular in power inspection operation and maintenance, such as: high-voltage line inspection UAV , substation inspection robot, 10kV room track inspection robot, AR smart wearable inspection equipment, etc. [0003] At present, in the field of power equipment image recognition, there are two problems: [0004] Problem 1. A large amount of image data collected by inspection is wasted [0005] At present, electric power intelligent inspection equipment will conduc...

AI Technical Summary

Problems solved by technology

[0004] Problem 1. A large amount of image data collected by inspection is wasted

Part of the intelligent inspection equipment uses local direct image recognition and detection, and does not save the inspection pictures; another method is to temporarily store them in the inspection background system, but it cannot be permanently stored (background server hard disk space factors, need to regularly clean up old pictures )

[0006] In short, a large amount of image data of power equipment cannot be effectively saved and recorded, resulting in a waste of a large amount of historical inspection operation and maintenance image data

[0007] Question 2. The detection department does not have a quantitative criterion for the image recognition of power equipment

[0011] 1. A large amount of power equipment image data cannot be effectively saved and recorded, resulting in a large amount of waste of historical inspection operation and maintenance image data

[0012] 2. Inspection departments such as the Electric Power Research Institute are relatively general in the detection of machine vision smart terminals, and cannot realize the visual function and accuracy detection of data quantification

Inspection personnel cannot know the prediction value of the inspection robot for these target recognition, cannot judge its accuracy, and cannot make requirements for the robot's image recognition optimization, so it is impossible to efficiently and quickly promote product improvement

Images