SMT大数据的产品质量预测办法和实现研究

SMT Big Data Product Quality Prediction Method and Realization Research

近些年,表面贴装技术(SMT)的自动化、信息化水平有很大提升,生产出的产品性能更优异。SMT生产工艺复杂,各道工序实施阶段均可能出现质量缺陷,若在生产线上不能尽早地洞察、处理这些缺陷,则很可能降低产品质量,严重时会造成整块印制板报废,增加产品的生产费用。文章从大数据技术角度预测基于SMT下锡膏印刷过程的产品质量,预测阶段设定了时间窗口,动态更新数据,最后创建了时间序列数据包,较明显地提升了模型预测的正确率。

In recent years,the automation and information level of surface mount technology(SMT)has been greatly improved,and products with better performance can be produced.The SMT production process is complex,and quality defects may appear in the implementation stages of each process.If these defects cannot be detected and dealt with as soon as possible on the production line,it is likely to reduce product quality,and in severe cases,the entire printed circuit board will be scrapped and the cost of products will be increased.Based on the perspective of big data technology,this article predicts the product quality of the solder paste printing process based on SMT.The prediction stage sets a time window T,dynamically updates the data,and finally builds a time series data package,which significantly improves the accuracy of the model prediction rate.