您的位置：首页 > 论文页面
基于分层技术的快速最大方差展开算法及其在过程监测中的应用
发表时间：2018-08-31 浏览量：1148 下载量：155
| 全部作者： | 魏驰航，宋执环，陈荣辉 |
| 作者单位： | 浙江大学控制科学与工程学院；中原大学工学院 |
| 摘 要： | 提出一种针对大规模数据的基于分层技术的快速最大方差展开（fast maximum variance unfolding，FMVU）算法。该算法通过分层技术求取近似核来显著减小计算复杂度和空间要求，同时最小化所牺牲的准确性。数据的局部特征和稀疏性保证了FMVU分层技术的有效实施。本文给出了FMVU算法的数学结构，然后定量地推导了精确性、计算复杂度和空间要求。通过一个数学例子和连续搅拌釜式加热器（continuous stirred tank heater，CSTH）过程验证了所提算法的可行性与有效性。 |
| 关 键 词： | 自动控制技术；过程监测；快速最大方差展开；分层技术；大规模数据 |
| Title： | Fast maximum variance unfolding algorithm and its application for process monitoring based on hierarchical technique |
| Author： | WEI Chihang, SONG Zhihuan, CHEN Junghui |
| Organization： | College of Control Science and Engineering, Zhejiang University; College of Engineering, Chung Yuan Christian University |
| Abstract： | In this paper, a learning framework for fast maximum variance unfolding (FMVU) algorithm based on hierarchical technique is proposed to learn a scalable approximate kernel that helps to reduce computational complexity and storage requirements for large-scale datasets, as well as minimize the sacrificed accuracy. The hierarchical construction of FMVU from the collected data is achieved by the good localization characteristics and sparsity of data. The mathematical framework for the development of FMVU algorithm and quantitative derivation on the accuracy, computational complexity as well as storage requirements are presented in this paper. The feasibility and efficiency of the proposed method is illustrated through a numerical case and the continuous stirred tank heater (CSTH) process. |
| Key words： | autocontrol technology; process monitoring; fast maximum variance unfolding; hierarchical technique; large-scale dataset |
| 发表期数： | 2018年8月第16期 |
| 引用格式： | 魏驰航，宋执环，陈荣辉. 基于分层技术的快速最大方差展开算法及其在过程监测中的应用[J]. 中国科技论文在线精品论文，2018，11（16）：1651-1658. |
2
评论数 0请您登录
暂无评论
下载全文
多维论文