Research On Product Quality Control Of G Enterprise Based On SMT Big Data Analysis

Surface mount technology(SMT)is the most critical process in manufacturing(Electronics),which has a direct impact on product quality.With the continuous upgrading of the electronic industry,the required assembly components tend to be miniaturized and integrated,and the corresponding PCB circuit board design is becoming more and more complex and exquisite,which has more stringent requirements for SMT core technology solder paste printing.If any small defects in the process can not be found in time and accurately,it may reduce the product qualification rate and increase the product life Production cost.SMT production line detects the solder paste printing link by adjusting the upper and lower thresholds of SPI solder paste detector,so as to reduce the product defects caused by printing defects.On the one hand,the setting of the upper and lower limits of the current SPI detector relies too much on human experience,and the analysis of various factors involved in the production process is insufficient.It is difficult to effectively detect the quality of the printed solder paste,so that the products with printing defects flow into the next process.On the other hand,all kinds of testing equipment on SMT production line generate massive data at all times,but only a few hidden information behind the data are applied,most of the data is not used for mining,which results in "data waste".Therefore,from the perspective of SMT big data,using industrial big data driven modeling technology,this paper proposes a new SPI threshold setting method to achieve the purpose of product quality control.The main contents and innovations of this paper are as follows.(1)The SMT production process of G enterprise and SMT big data resources are comprehensively described.Based on this,a new model of SPI upper and lower threshold estimation is designed.Firstly,the SMT packet are constructed;Secondly,the upper and lower threshold estimation dynamic models of SPI are designed;Finally,the model is optimized.The distribution characteristics of SPI data are determined by mechanism analysis and nonparametric density estimation.(2)Firstly,smote over sampling technology and under sampling technology are used to solve the problem of unbalanced distribution of SMT data categories;Secondly,Gaussian kernel density estimation method is used to estimate the probability density of three types of data in SMT packets;Thirdly,Bayes decision classification criterion is used to build the SPI detection parameter threshold estimation model with decision variables as the upper and lower limits of SPI threshold estimation and the goal of minimizing the misjudgment rate and missed judgment rate of SPI;Finally,Genetic algorithm(GA)is used to optimize the model to obtain the optimal upper and lower thresholds of SPI.(3)The SPI upper and lower thresholds obtained by model optimization are applied to SMT pipeline of G enterprise.Through the analysis of examples before and after SPI threshold optimization,it is found that the tin bonding rate of SMT products can be reduced by 0.04673% by appropriately reducing the upper threshold value(from 200% to 142.56%);Increasing the lower threshold value(from-65% to-73.74%)can reduce the solder void ratio by 0.01947%.So as to reduce the production cost of SMT products,improve the pass rate of SMT products,and provide effective help for SMT product quality control.