Research On Detection System Of Welding Defects Of Copper Tube Connector Based On Machine Vision

The copper tube connector is one of the widely used parts in air conditioning,which is mainly used for connecting the different diameter pipelines in the air conditioner.The copper pipe connector is composed of copper and stainless steel ring by welding process.Because of the material,the solder and the welding process and so on,the welding process often produces welding defects.At present,for the detection of the defects,the company mainly adopts the method of artificial visual inspection.However,the method has low precision and long time,furthermore,it is dependent on the subjective judgment of human,which can’t meet the requirements of enterprise automation testing.Therefore,it is of practical significance to design a set of welding defect detection device for copper tube.In this paper,the automatic detection of welding defects of copper tube is studied as the research object,main research contents are as follows:(1)In the analysis of the welding defects of the copper tube,three kinds of defects are obtained,which are the defects of the weld zone,the hole and the solder.According to the actual requirement of the production of copper tube,the technical index of the detection system is as follows: the defect detection accuracy is not less than 90%,the average detection time of a single purple copper tube is less than 3 seconds.(2)The defect detection scheme based on machine vision technology is selected,which is composed of a feeding unit,a detection unit and a sorting unit.At the same time,the overall working process of the device is designed.(3)The detection unit is designed.The hardware part includes image acquisition module,control module,case mechanism and PC machine.The software part includes upper computer software and lower computer software.(4)The welding defect detection algorithm is designed,A threshold method is used to obtain the image of the area containing the welding zone and the solder residue area.The area features zone welding and multiscale edge Fourier descriptors are combined to extract weld zone.(5)Technical indicators of the test equipment was tested.The defect detection accuracy rate is 92.2% and the average detection time of a single copper tube connector is 1.53 seconds,which meet the design requirements.