基于云计算的基板玻璃缺陷神经网络分类模型研究  

Study on Neural Network Classification Model of Substrate Glass Defects Based on Cloud Computing

作　　者：李青 周波 

机构地区：[1]东旭集团有限公司,石家庄050021 [2]平板显示玻璃技术和装备国家工程实验室,石家庄050035

出　　处：《计算机与数字工程》2017年第7期1373-1376,共4页Computer & Digital Engineering

基　　金：国家科技支撑计划(编号:2013BAE03B02)资助

摘　　要：基板玻璃缺陷的种类识别,是调整优化生产工艺的重要依据。论文研究了云计算的基本模式和原理,结合液晶基板玻璃生产中的实际问题,设计了一种基于云计算基板玻璃缺陷神经网络分类模型,对模型中处理工作的分配,传输数据包、调度机、云计算服务器进行了简要研究,该模型具有提升神经网络收敛速度;实现资源共享,提升生产效率;冗余计算和算法热升级;降低维护难度等优势。对于基板玻璃生厂商具有一定的参考意义。Identify types of glass substrate defects is an important basis for the adjustment and optimization of the production process. In this paper basic mode and principle of cloud computing are studied, combined with practical problems of LCD glass sub-strate production, based on cloud computing glass substrate defects classify model of neural network is designed, the allocation mod-el in processing and transmission of data packets, machine scheduling, cloud computing server are studied briefly, the model has to enhance the convergence speed of neural network, the sharing of resources is realized, production efficiency, is improved redun-dant computation and algorithm of thermally are upgraded, the maintenance difficulty and other advantages are reduced. For the glass substrate manufacturer it has a certain reference value.

关 键 词：基板玻璃 缺陷分类 云计算 神经网络 图像处理 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]