Insulator defect detection method based on machine learning

Abstract

The invention discloses an insulator defect detection method based on machine learning, comprising: converting an insulator picture into a plurality of signal values, taking the axial direction of the insulator as the row direction of the picture, and taking the direction perpendicular to the row direction as the column direction, and the picture The row direction of the image is the time direction of the signal, and the gray value of the column direction of the picture is the signal strength value; the identification model is set; the signal value is input into the identification model, and the identification model outputs the identification result. The insulator defect detection method based on machine learning disclosed by the present invention has a high recognition rate, especially for insulator damage, flashover, crack and insulator pollution, and the recognition effect is better.

Description

technical field [0001] The invention relates to an insulator defect detection method, in particular to a machine learning-based insulator defect detection method, which belongs to the technical field of railway detection. Background technique [0002] Insulators are an important part of railway transmission lines, and whether they are in good condition directly determines the stability of high-speed rail. [0003] As the service life of insulators increases, some insulators are bound to be damaged. Therefore, it is necessary to inspect the insulators to find and solve problems in time. [0004] The traditional detection method is human observation, which has a large workload, low accuracy, slow detection speed and huge cost. [0005] In the prior art, there is a method of using a high-definition camera to take photos of insulators and then judge the photos. However, because the surface of the insulator has ceramics or similar highly reflective materials, the quality of the ...

AI Technical Summary

Problems solved by technology

[0004] The traditional detection method is human observation, which has a large workload, low accuracy, slow detection speed and huge cost

[0005] In the prior art, there is a method of using a high-definition camera to take photos of insulators and then judge the photos. However, because the surface of the insulator has ceramics or similar highly reflective materials, the quality of the photos is seriously affected, resulting in a low recognition rate when using traditional images to identify insulator defects.

[0006] In the prior art, there is also a way to process the insulator graph first and then identify it. For example, the patent No. "CN 109753929" is an invention patent "a method for image recognition of high-speed rail insulators combined with a library". Although this method can Improve the recognition rate to a certain extent, but still cannot completely solve the problem of light interference caused by reflections

[0007] In addition, there are methods based on insulator profile recognition in pictures, such as the patent "CN 103247044" patent "Bad state detection method based on the curve and point singularity characteristics of high-speed rail catenary insulators", this method is mainly for insulators In addition, it has high requirements for photos and must be taken at night to reduce light interference. The adaptability is weak, and the recognition accuracy is still not high.

Images