Research On The Defect Prediction Method For The Smt Solder Paste Printing Process

Surface mount technology(SMT)is widely used in electronic assembly industry.It mainly includes three processes: solder paste printing,component mounting and reflow soldering.At present,the layout of printed circuit board(PCB)is becoming more and more complex and sophisticated,which brings great challenges to SMT solder paste printing.Solder paste printing is affected by many factors,so the quality of printing is difficult to guarantee.With the intellectualization and digitalization of SMT production line processing equipment,a large amount of data generated during SMT production can be automatically collected.Mining useful information from data to predict printing defects is an effective way to improve the quality of solder paste printing.Based on the solder paste printing data collected from SMT production line,this paper comprehensively analyses the various factors affecting solder paste printing,then obtains the influencing factors of solder paste printing defects,and on this basis,constructs and optimizes a solder paste printing defects prediction model to realize the advance prediction of printing defects,so as to improve the quality of solder paste printing.The main contents of this paper are as follows:(1)The overall framework of SMT solder paste printing defect prediction is designed.On the basis of fully understanding SMT solder paste printing mechanism and combing solder paste printing data resources,The overall framework of SMT solder paste printing defect prediction is designed,which includes three stages: data acquisition and exploration,influencing factors analysis,model construction and optimization.The framework is used to guide the implementation of solder paste printing defect prediction.(2)The influencing factors of SMT solder paste printing defects are analyzed.In view of the variety and correlation of solder paste printing parameters,it is difficult to determine the influencing factors of solder paste printing defects.To solve the difficult problem,a multilayer feature selection method based on improved ReliefF is proposed to analyze the influencing factors of solder paste printing defects.The influencing factors of solder paste printing defects are preliminarily screened by improved ReliefF algorithm,and then Pearson correlation analysis is carried out to eliminate the redundant influencing factors.The evaluation shows that the proposed method is superior to other typical feature selection methods.(3)The mixed sampling method for SMT solder paste printing data is studied.In view of the unbalanced proportion of qualified printing data and defective printing data,the prediction model of solder paste printing defects is affected.In this paper,a hybrid sampling method based on clustering sampling and synthetic minority oversampling technique(SMOTE)is proposed to sample solder paste printing data.Clustering sampling is used to undersample qualified printing data,and SMOTE algorithm is used to oversample printing defect data.(4)SMT solder paste printing defect prediction model is constructed and optimized.According to the influencing factors of solder paste printing defects and the solder paste printing data after mixed sampling,the solder paste printing defect prediction model based on extreme gradient boosting(XGBoost)is constructed.and genetic algorithm is used to optimize the model,the advance prediction of solder paste printing defects is achieved.Case analysis of data collected by SMT production line show that the proposed analysis idea and method can accurately predict the printing defects of solder paste.