Automated Weld Inspection And Rating System

With the development of equipment manufacturing,people pay more and more attention to the quality and safety of products,and the requirements for automation and intelligence are also higher and higher.According to the requirements of nuclear power equipment manufacturing quality assurance system,it is necessary to use X-ray as a non-destructive testing method to detect the welding quality of cylinder workpiece.At present,manual operation is used to complete film replacement and shooting.The detection quality is low and the detection efficiency is not high.This problem is common in other industries that need to use X-ray inspection.This paper proposes to design a scheme to realize automatic weld detection,in order to maximize the detection speed,improve the detection level from the source,and effectively meet the application requirements of various scenarios.Firstly,this paper analyzes the background,research status,purpose and significance of automatic weld defect detection,as well as the actual process of weld detection,and makes the overall design and planning of the system.Secondly,the work flow of the system is discussed.Based on the overall design plan,the hardware design is completed according to the specific indicators to be achieved according to the actual needs,and the operation characteristics of the system are clarified,and the detection principle of weld image is introduced in the detection design part.Thirdly,the software part of the system is designed.Lab VIEW is used to program the upper computer,and the information input and control are carried out based on the human-computer interface with strong interaction,so as to achieve the purpose of monitoring the system running state comprehensively.The lower computer uses PLC,on the one hand,it controls the mechanical structure,on the other hand,it complements the missing function of the upper computer.Finally,the image processing part mainly uses the semantic segmentation model based on U-net algorithm,and improves the network through migration learning.Integrating Res Net34 network into U-net network not only has fast detection speed,but also has high detection precision.It can help enterprises to complete automatic flaw detection tasks of welding seams,generate inspection welding quality rating results,and record the types and locations of defects in the generated data,which can help workers to quickly deal with unqualified products and improve production efficiency.Based on the research of the above-mentioned software and hardware systems,this paper has successfully realized the design of an automated weld inspection and rating system.This detection system and method can be used for reference for similar problems in related technical fields.