基于改进DNN网络的电装产线质量预测方法  

Improved DNN Network Method for SMT Production Lines Quality Prediction

作　　者：许政 阮西玥 XU Zheng;RUAN Xi-yue(AVICAS Generic Technology Co.,Ltd.,Yangzhou 225000,China)

机构地区：[1]中航机载系统共性技术有限公司,江苏扬州225000

出　　处：《航空计算技术》2024年第4期125-129,134,共6页Aeronautical Computing Technique

基　　金：MJ专项科研项目资助(MJZ4-XXXX)。

摘　　要：随着航空机载电装模块功能多元化、尺寸精细化、器件复杂化程度的不断提升,对SMT产线电装质量提出了新的挑战,目前SMT产线生产过程普遍存在产品质检数据关联性较差,产品质量分析滞后及预测性差等问题,而传统的统计分析方法无法有效提取海量无序数据中的知识和规律,提出一种基于深度学习的电装质量预测方法。首先,构建质量评价方法,确定电装质量的影响因素;其次,利用主成分分析法(Principal Components Analysis,PCA)对质量数据进行预处理,剔除非相关特征;然后,引入DNN网络,构建电装质量预测模型,利用BFO-PSO优化算法搜寻网络的最优隐含层层数及节点数;最后,通过航空电装产线实际制造数据进行仿真测试,验证了所提出方法的有效性和科学性。With the continuous improvement of function diversification,size refinement and device complexity of airborne electrical assembly modules,new challenges have been posed to the quality of electrical assembly of SMT production lines.At present,the manufacture process of SMT production lines generally has problems such as weak correlation of product quality inspection data,lagging product quality analysis and weak predictability.However,the traditional statistical analysis method cannot effectively extract knowledge and rules from massive disordered data,this paper proposes a deep learning-based electronic assembly quality prediction method.To begin with,the quality evaluation method is constructed to determine the influencing factors of electric assembly quality.Then,use Principal Components Analysis(PCA)to preprocess the quality data and eliminate the non-relevant features.Furthermore,the DNN network is introduced to construct the quality prediction model,and BFO-PSO optimization algorithm is used to search the optimal hidden layer number and node number of DNN network.Finally,based on the actual manufacturing data of the aviation SMT production lines,the module quality prediction model is simulated and tested to verify the effectiveness and scientificity of the proposed method.

关 键 词：SMT产线 电装质量预测 BFO-PSO优化算法 DNN网络 智能制造 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]