深度学习在汽车制造物流规划工作中的应用  

Application of Deep Learning in Scheduling of Automobile Production Logistics

作　　者：钱丰 梅剑平 潘荣胜 

机构地区：[1]上汽大众汽车有限公司计划物流部,上海201805

出　　处：《物流技术》2017年第12期84-89,共6页Logistics Technology

摘　　要：针对汽车制造物流规划中体积参数的预估问题,应用深度神经网络学习方法进行了输出数据损失函数设计,输入数据采集与增强和神经网络的构建。按此进行的实验结果显示此方法可大幅提高零件体积预估输出的准确性与稳定性。同时也为深度学习在物流规划日常工作中的应用提供了途径。In this paper, in view of the issue of estimation of volume parameter in the scheduling of automobile production logistics, we used the neural network deep learning process to design the output data loss function and build the neural network model for input data collection and augmentation and according to the result of a numerical example using the model, demonstrated its capability to greatly enhance the accuracy and stability of the estimate output of spare part volume. At last, we pointed out the path for applying deep learning in the daily logistics planning activities.

关 键 词：深度学习 神经网络 物流规划 汽车制造 

分 类 号：F426.471[经济管理—产业经济] U466[机械工程—车辆工程]