Lithium Battery Pole Piece Defect Detection Technology Based On Machine Vision

Lithium battery is a kind of high-density and long-life environmentally friendly power battery,which is deeply favored by the modern power battery market.The pole piece is the basis of the battery.Any defect on the pole piece will seriously affect the quality and safety performance of the lithium battery.Therefore,in the process of producing lithium batteries,it is necessary to carry out defect detection,and adjust the pole piece production machine and Craft.At present,most domestic lithium battery manufacturers still use traditional manual detection of defects,which is low in efficiency and easily affected by the subjective factors of workers,resulting in missed and wrong detections.On the one hand,the flow of unqualified pole pieces to the next process will directly cause the production of lithium batteries to fail to meet the standards;on the other hand,companies have put forward higher requirements for defect detection efficiency and accuracy.Aiming at the problems of low efficiency in manual detection of pole pieces and easy to miss and wrong detection,this paper uses machine vision technology to detect defects in lithium battery pole pieces.This method can improve work efficiency and accuracy,and reduce enterprise production and operation costs.This paper takes lithium battery pole piece as the research object to carry out research work.The defects of lithium battery pole pieces mainly include surface scratches,bubbles and holes.This paper designs a method for detecting defects of lithium battery pole piece based on machine vision technology for these three kinds of defects,and realizes the online continuous detection of lithium battery pole piece defects.The main work content is as follows:First,according to the working requirements and working environment,choose the camera,lens,light source and lighting method,build the pole piece image acquisition platform,and complete the camera calibration calculation.Secondly,in order to improve the quality of the pole piece image,extract the pole piece coating area,improve the brightness and contrast of the area,use the Otsu method to binarize the image and then use morphological processing for filtering operations.On the basis of the original algorithm,GPU parallel processing is used,Otsu method and morphological processing are optimized,and the processing time of the two operations is shortened.After experimental verification,the speedup ratio is about 1.45 and 2.12 respectively.Then use an improved Canny edge detection algorithm to detect the contour of the defect,and use the image gradient value to calculate the geometric characteristics of the defect.Finally,the HOG features and geometric features are combined to train the support vector machine.According to the trained SVM model,the defective images are classified and stored in the database,and the final classification accuracy rate is 94.67%.The experimental results prove that the system can effectively increase the detection rate and realize the detection and classification of three kinds of defects,which has certain feasibility.