摘要
针对制造过程的在线质量预测的实时性问题,提出了一种融合云加端的在线质量预测架构。该架构在云加端提出一种基于遗传算法(GA)参数优化的隐含层节点自适应增长极端学习机(AGELM)方法,建立了优化的产品质量预测模型。同时,该架构在终端改进了k-means方法并将其应用于在线质量数据流聚类,并将聚类中心序列输入产品质量预测模型,预测产品的质量。通过点焊过程的实验表明该产品质量预测模型方法实时性较BP神经网络和贝叶斯方法有较大优势,能应用于当前制造过程的在线质量预测。
According to real-time problems of online quality prediction in manufacturing,this paper proposed a framework of online quality prediction of manufactured products based on cloud computing and terminal computing. In the framework,a hidden layer node adaptive growth extreme learning machine(AG-ELM)method based on parameter optimization of the genetic algorithm(GA) is proposed and an optimized model of product quality prediction is established in cloud computing. The method of K-means is improved to cluster the online quality data stream and the sequence of clustering centers is input into the model of product quality prediction to predict the quality of the product in terminal computing. The experiment of spot welding showed that a framework of online quality prediction this paper proposed was super to BP neural network and Bayesian and could be applied to the online quality prediction of manufacturing process.
作者
唐向红
易向华
陆见光
元宁
刘国凯
TANG Xiang-hong YI Xiang-huala LU Jian-guang YUAN Ning LIU Guo-kai(a. Key Laboratory of Advanced Manufacturing Technology, Minist ry of Educat ion b. School of Mechanical Engineering, Guizhou University, Guiyang 550025, China 2. Guizhou Provincial Key Laboratory of Public Big Data, Guiyang 550025 , Chin)
出处
《组合机床与自动化加工技术》
北大核心
2017年第5期64-68,共5页
Modular Machine Tool & Automatic Manufacturing Technique
基金
贵州省重大科技专项(黔科合重大专项字[2013]6019
黔科合重大专项字[2012]6018)
贵州省基础研究重大项目(黔科合JZ字[2014]2001)
