Research On Optimization Method Of Key Process Parameters Of SMT Production Line Based On Artificial Intelligence

With the increasing digitization of the Surface Mount Technology(SMT)production line,the use of massive data in the production process of the SMT production line to optimize key process parameters is an effective means to improve the quality of SMT products.There are a large number of process parameters in the SMT production line process flow.These parameters have an inseparable and complex relationship with the quality of the final product.Therefore,it is particularly important to optimize the key process parameters of the SMT production line.At present,the optimization of process parameters has the following problems in the actual production of SMT production lines: Low efficiency as the parameters could meet the requirements only with a larege quantity of single board trial production adjustments;Underutilization of the data in the production process;Insufficient exploitation for the relationship between process parameters and product quality.In recent years,artificial intelligence technology has performed well in various fields,providing a new idea for the optimization of key process parameters of the SMT production line.Therefore,this thesis constructs a new process parameter optimization method based on related technologies such as deep learning and reinforcement learning in artificial intelligence.The main contents are as follows:(1)This thesis sorts out the SMT production line process flow,the solder paste printing process and the reflow soldering process.With detailed analysis about the mechanism of the solder paste printing process and the reflow soldering process and data resource preprocessing,the overall framework for the optimization of process parameters based on artificial intelligence is proposed.Therer are mainly two parts in the optimization method of key process parameters in SMT production line: SMT production line printing process parameter optimization method and SMT production line reflow soldering process parameter optimization method.(2)Realize the optimization of solder paste printing process parameters based on reinforcement learning and intelligent optimization algorithms.First,in view of the high dimensional characteristics of the solder paste printing process,the influencing factors of printing quality were extracted from the data resources in the printing stage of SMT production line by using feature selection based on XGBoost and feature reconstruction based on FC-PCA;then,according to the characteristics of the quality of the printed pads,a printing quality prediction model based on the improved DDPG algorithm is proposed to fully explore the relationship between influencing factors and printing quality.Finally,in view of the problem that the volume and area of the solder paste contribute to the different weights of the solder paste printing quality,the adaptive weight genetic algorithm is used to optimize the printing process parameters,which avoids the blindness of manually setting the optimization target weight.(3)This thesis realize the optimization of reflow soldering process parameters based on the integrated deep neural network and intelligent optimization algorithm.Firstly,the traditional BP neural network has difficult in accurately mapping the characteristics of the reflow soldering process,the reflow soldering temperature curve prediction model is constructed by integrating multiple deep neural networks optimized by EGA to accurately map the relationship between the reflow soldering process parameters and the temperature zones of the reflow soldering,and improve the prediction accuracy.Then,in view of the fact that it is difficult to select Pareto solutions for the large dimensions of the reflow temperature curve,the NSGA-? algorithm adopts a new selection mechanism to optimize the reflow process parameters,which effectively reduces the calculation cost.Based on the above research content,an example analysis of the optimization method of the key process parameters of the SMT production line has been completed,and the production line production verification has proved the effectiveness of the method proposed in this thesis.