Research On Defect Detection Method Of Industrial Control Circuit Board Based On Deep Learning

Industrial control circuit board is the core parts of engineering control,in the industrial production line of industrial control circuit board,defect detection is an extremely important part,because of the current industrial control circuit board surface line,size and shape are very different due to the different functions,so the detection of industrial control circuit board defects has been the algorithm accuracy is not high,there are missing,false detection.Aiming at the limitations of current defect detection methods,this paper proposes an improved YOLOv5 industrial circuit board defect detection method based on deep learning.An improved YOLOv5 target detection algorithm is proposed by analyzing the research achievements and existing problems in the field of industrial control circuit board detection at home and abroad.The attention mechanism module was added to the backbone network of the original detection model to improve the detection accuracy by improving the feature extraction ability of small targets.The improvement effects of SENet and CBAM were compared and analyzed.The K-Means++ clustering algorithm was used to improve the matching degree between the predicted box and the real box.On the basis of feature extraction fusion module,ASFF multi-scale fusion module is added to further strengthen the feature fusion ability of YOLOv5 detection model for images of different scales,prevent the problem of information loss of different feature layers of industrial control circuit boards,and improve the overall detection performance and generalization ability of defect detection model.A visual system for defect detection of industrial control circuit board is designed.The system mainly includes the selection module of detection model,input type selection module,Io U and confidence adjustment module,statistical preservation module of detection results and main window module of defect detection of industrial control circuit board,which can realize the visualization of defect detection of industrial control circuit board.The experiment compares the detection effect of the improved YOLOv5 model with the original model and other mainstream target detection models.The results show that the detection speed and accuracy of the improved YOLOv5 model are significantly improved compared with other models.The average recognition accuracy can reach96.03%,and the detection speed takes about 0.0127 s per image,which verifies the effectiveness of the improved algorithm proposed in this paper and provides a new method for defect detection in the field of industrial control circuit board production.