一种改进的模糊PLS模型在软测量中的应用  

Application of a Modified Fuzzy Partial Least Squares Model in Soft Measurement

作　　者：张先林[1] 于佐军[1] 

机构地区：[1]中国石油大学(华东)信息与控制工程学院,山东东营257061

出　　处：《控制工程》2008年第S2期118-120,160,共4页Control Engineering of China

摘　　要：针对石油化工过程的特点,提出一种改进的模糊部分最小二乘的建模算法。该算法先用减法-模糊C均值聚类进行模糊结构辨识,然后用高斯核函数实现非线性PLS并且用其进行后件参数辨识,即得到多个子模型,再将各模型输出的数据进行隶属度加权求和得到最终的估计输出。最后,将该方法应用于航煤干点的估计,仿真结果表明该算法更有效,预测精度更高。A modified fuzzy partial least squares(PLS) model is presented for the characteristics of petrochemical process.In this method,the fuzzy structure is made by subtractive-fuzzy C-means clustering firstly,then the consequence parameter identification is obtained by a modified nonlinear PLS which is got by Gaussian kernel function,and some sub-models are gained.And then the degrees of membership are used for combing several sub-models to obtain the final result.Finally,this method is applied to the estimation ...

关 键 词：模糊部分最小二乘 模糊C均值聚类 减法聚类 软测量 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]