文献类型：期刊文献

中文题名：最小二乘支持向量机的参数优化及其应用

英文题名：Parameters Optimization of LS-SVM and Its Application

作者：陈帅[1];朱建宁[2];潘俊[2];侍洪波[1]

机构：[1]华东理工大学自动化研究所,上海200237;[2]上海焦化有限公司,上海200241

年份：2008

卷号：34

期号：2

起止页码：278

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：20081911245050);Scopus;北大核心:【北大核心2004】；CSCD:【CSCD2011_2012】；

基金：上海市曙光计划资助项目(03SG26)

语种：中文

中文关键词：最小二乘支持向量机;进化类算法;参数优化;遗传算法;粒子群算法;BP神经网络;德士古气化炉;软测量建模

外文关键词：LS-SVM; evolutional algorithm; parameters optimization; GA; PSO; BP neural network; Texaco gasifier; soft sensor modeling

摘要：针对最小二乘支持向量机的多参数带来的参数寻优问题,将进化算法(遗传算法和PSO算法)应用其中,通过Sinc函数的测试,成功地实现了多参数的联合优化;将这一方法应用到德士古炉温软测量建模中,采用来自工业现场的实测数据进行仿真,将两种方法的仿真结果与常用的BP神经网络进行比较,可以看出两种算法都较好地解决了最小二乘支持向量机的参数优化问题。
LS-SVM has several parameters needed to be optimized, so it is difficult to do all these optimization. Evolutional algorithm, such as genetic algorithm and PSO algorithm is used to solve this problem. Through the testing of the function Sine we can see that it can realize parameters optimization of LS- SVM. Then this method is used in the soft sensor modeling for temperature measurement of Texaco gasifier. By comparing the results using these two algorithms and comparing those results with BP neural network which is often used, we can see that they can both solve this problem of least squares support vector machines and obtain good performances.