详细信息
基于贝叶斯学习的关联向量机及其在软测量中的应用 ( EI收录)
Relevance Vector Machine Based on Bayesian Learning and Its Application in Soft Sensing
文献类型:期刊文献
中文题名:基于贝叶斯学习的关联向量机及其在软测量中的应用
英文题名:Relevance Vector Machine Based on Bayesian Learning and Its Application in Soft Sensing
作者:陈佳[1];颜学峰[1];钱锋[2]
机构:[1]华东理工大学自动化系,上海200237;[2]华东理工大学化学工程联合国家重点实验室
年份:2007
卷号:33
期号:1
起止页码:115
中文期刊名:华东理工大学学报(自然科学版)
外文期刊名:Journal of East China University of Science and Technology
收录:CSTPCD;;EI(收录号:20071510544974);Scopus;北大核心:【北大核心2004】;CSCD:【CSCD2011_2012】;
基金:国家自然科学基金(20506003);上海科技启明星项目(04QMX1433);教育部科学技术研究重点项目(106073);国家973计划(2002CB3122000);国家杰出青年科学基金(60625302);国家863计划项目(20060104Z1081);上海市科委重大基础研究(05DJ14002);上海科委科技攻关项目(04DZ11010;05DZ11C02)
语种:中文
中文关键词:稀疏贝叶斯;关联向量机;软测量;溶剂脱水塔
外文关键词:sparse Bayesian; relevance vector machine; soft sensing; solvent dehydration tower
摘要:介绍了一种与支持向量机(SVM)函数形式相同的稀疏概率模型——关联向量机(RVM),其训练是在贝叶斯框架下进行的,在处理具有噪声的函数回归时,RVM具有很出色的性能。与SVM相比不仅解更稀疏,而且无需调整模型参数,核函数选择也不受限制。将RVM应用于PTA装置溶剂脱水塔塔顶塔底组分软测量建模,仿真结果表明:该方法预测精度较高,具有一定的应用价值。
Based on Bayesian learning, a sparse probabilistic model, termed as "relevance vector machine'(RVM), is introduced, which has the same functional form as support vector machine (SVM). RVM has an excellent ability in dealing with functional regression with noise. Compared with SVM, RVM has sparser solutions. Besides, RVM doesn't need turning any parameters of model, and has not any restriction on the selection of kernel function. Finally, RVM was utilized to develop a soft sensor of pure terephthalic acid (PTA) solvent dehydration tower. The simulation results show that the proposed method has higher precision of prediction.
参考文献:
正在载入数据...
