Research On Welding Defect Detection Of Square Battery Pack Based On Machine Vision

With the rapid development of modern new energy vehicles,the demand for electric vehicles is increasing.The square battery pack has become the darling of the market due to its advantages of high energy density,lightweight,recycling,environmental protection,and strong endurance.The welding automation of the pole poles of the square battery pack is getting higher and higher,but there will be various welding defects in the welding process of the pole poles produced.For the current detection of welding defects in the pole poles of square battery packs,traditional manual visual inspection methods are still used in factory inspection,which affects the detection efficiency and continuity of work,and there will be errors in the classification of detection defects.To this end,this article applies Machine Vision to the detection of welding defects on the poles of the square battery pack to improve detection accuracy and efficiency.The important and difficult point of detection is how to extract the welding area in the pole image of the square battery pack and select appropriate features for classification and recognition.The main content of this article includes the following four aspects:First,it mainly introduces the construction of the welding defect detection platform of the pole poles of the square battery pack and the design of the overall scheme.The detection platform is very important for defect detection.According to the needs of detection,the corresponding design of the image acquisition platform is completed,including the optical system,light source system,and motion control platform.The experimental platform can complete the image acquisition of the pole module.Secondly,related research is done on the preprocessing algorithm of pole welding images.Including image filtering,enhancement,and segmentation methods.The median filter algorithm,linear enhancement transformation method,and maximum between-class variance segmentation method are used to effectively remove salt and pepper noise and impulse noise in the image,enhance the contrast of the welding image of the pole to be tested,and improve the efficiency of subsequent defect detection.Then,because of the wide variety of pole welding defects and inconspicuous features,two defect detection,and recognition algorithms are used.Among them,Halcon’s operator is called for defect detection.This algorithm can identify whether all samples are defective,and quickly and accurately detect most types of defects such as weld penetration,thin welds,unwelded,and virtual welds.However,it is easy to blow up the welding.The point is misidentified as welding through.Therefore,after analysis,the Ray-shooting algorithm is used for classification and recognition of defect detection.The Ray-shooting model is around core designed to extract the pixel intensity features around the seed point.First,based on the characteristics of the welding image,the Otsu algorithm and morphological operations are used to extract the welding area.Then,use the distance transformation to get the center point of the welding area and generate the Ray-shooting model to extract the features at this point.Finally,the types of welding defects are classified based on the extracted features.Experiments have proved that this algorithm is highly robust,and can detect and classify defects such as false soldering,missing soldering,fried spots,and solder penetrations of the pole pads.The classification and recognition accuracy of defects on all test pictures is 100%.Finally,a set of terminal welding defect detection software platforms was developed using the Microsoft Visual Studio software platform.A PC operation interface is provided for the defect detection platform.