基于机器学习的回流焊焊点形貌预测

提出一种基于机器学习预测回流焊焊点形貌的方法,通过该方法建立一个针对钽电容回流焊焊点形貌的预测模型,该模型为现有实验方式提供了新的思路。通过峰值温度、降温速率和焊膏厚度3种影响因素以及焊点厚度、焊点宽度和焊料爬高3种评价焊点形貌的评价标准,分别基于BPNN和LightGBM算法建立钽电容回流焊焊点形貌预测模型。对比实验证明,通过LightGBM算法建立的预测模型优于通过BPNN建立的预测模型,并通过实际测试帮助实验人员减少实验次数,节约大量时间成本。

Topography Prediction of Reflow Solder Joints Based on Machine Learning

A method for predicting the topography of reflow solder joints based on machine learning is proposed.Through this method , a prediction model for the topography of reflow solder joints of tantalum capacitors is established , which provides a new idea for the existing experimental methods.Based on the three influencing factors of peak temperature , cooling rate and solder paste thickness , and three evaluation criteria for evaluating solder joint morphology , solder joint thickness , solder joint width and solder climb , the reflow soldering of tantalum capacitors was established based on BPNN and LightGBM algorithms respectively.For the point topography prediction model , the comparison experiment proves that the prediction model established by the LightGBM algorithm is better than the prediction model established by the BPNN , and helps the experimenter to reduce the number of experiments and save a lot of time and cost through actual testing.