Design And Implementation Of SMT Big Data Analysis Platform

Surface Mount Technology(SMT)is a key technology in the electronics manufacturing industry.With the miniaturization and complication of electronic products,more stringent requirements are imposed on the core process of SMT,solder paste printing.In the printing production practice,due to improper setting of printing parameters,it often leads to problems such as inaccurate printing position,many defective prints and long verification cycle of printing parameters.At present,the quality analysis methods of solder paste printing mainly include artificial experience,experimentation and traditional SPC(Statistical Process Control)technology.Faced with the massive data of SMT,these methods have problems such as low analysis efficiency,insufficient analysis accuracy,low utilization rate of SMT data and lack of effective verification means.In view of the above problems,this paper develops a big data analysis platform for the SMT industry,which can extract the hidden rules in the printing parameters through data analysis methods,help the quality analysts to find out the causes of printing problems in a short time;The visual simulation technology is used to construct the SPI detection simulation verification module of the SMT production line to verify the accuracy of the analysis results.The main research contents of the thesis are as follows:(1)Through the demand analysis of the SMT big data analysis platform,the platform is divided into data support layer,data storage computing layer,platform function layer and platform application layer according to the layering idea;and the technical route is formulated based on the functional layer of the platform.the functional module is designed.(2)According to the actual business requirements of solder paste printing parameters,the SMT big data processing and analysis process was designed,and the SMT production line data collection scheme and real-time data processing architecture were designed.According to the characteristics of SMT production line data,the K-means algorithm is used to detect the outliers of the data,and the accuracy of the method is verified.In addition,the recommendation model of the printing rules based on Spark is designed.The example is verified by parallelization.The efficiency of the Apriori algorithm is significantly improved.(3)Based on the 3D visualization simulation technology,the design and development of the SPI detection simulation verification module is completed.The simulation verification model of SPI detection based on GA-BP neural network is designed.The maximum average relative error of the model is 5.1374%,Precision requirements for solder paste printing quality;completed the construction of a virtual model of SMT production line equipment.The simulation verification module is used to verify the recommended 5 sets of printing parameters.The accuracy of the simulation verification results is above 95%,and the accuracy is high.It proves that the recommended model of solder paste printing parameters based on association rules has higher accuracy.(4)Based on the open source big data frameworks such as Hadoop and Spark,the development of SMT big data analysis platform was completed by using Java Web development related tools.