基于LSSVM的溶剂油分离建模方法研究  

Research on modeling methods for solvent oil separation based on LSSVM

作　　者：黄燕[1] 薄翠梅[1] 牛超[1] 

机构地区：[1]南京工业大学电气工程与控制科学学院,江苏南京211816

出　　处：《现代化工》2017年第2期190-193,195,共5页Modern Chemical Industry

基　　金：国家自然科学基金项目(61203020);江苏省自然科学基金项目(BK20141461)

摘　　要：针对传统支持向量机和单一模型建模的缺点,利用某炼油厂溶剂油分离过程中二侧线流量作为建模对象,对最小二乘支持向量机集成学习方法进行了研究。首先利用自适应系数加权模糊(AWFCM)聚类算法对训练样本进行聚类;然后对每一类数据使用最小二乘支持向量机建立子模型,并使用PLS合成函数得到最小二乘支持向量机集成模型;最后通过仿真实验来验证最小二乘支持向量机集成模型预测的精确性。结果表明,该算法在预测精度上有了较大的提高,对过程控制系统中分离效果的预测具有重要指导意义。As traditional support vector machine and single modeling have some inevitable shortcomings,an integrated learning method of least squares support vector machine（LSSVM） is studied by using two-side line flow in the solvent oil separation of a refinery as the modeling object. The adaptive coefficient weighted fuzzy（AWFCM） clustering algorithm has been used to cluster the training samples. LSSVM is used to establish sub-models for each category of data.The LSSVM integrated models is then obtained by using PLS synthesis function. Finally,simulation experiments are carried out to verify the accuracy of the LS-SVM integrated model. The results show that the proposed algorithm has a great improvement in prediction accuracy and has important guiding significance for the prediction of separation effect in process control system.

关 键 词：支持向量机 最小二乘法 聚类算法 溶剂油分离 软测量 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]