混合核函数稀疏LS-SVM软测量建模与应用

针对最小二乘支持向量机存在的稀疏性欠缺和单核函数局限性问题,本文提出一种基于混合核函数稀疏最小二乘支持向量机的软测量建模方法.该方法使用多项式核函数和RBF核函数线性加权构成混合核函数,兼顾最小二乘支持向量机的全局拟合能力与局部拟合能力,以矢量基学习作为稀疏解算法,改善最小二乘支持向量机的稀疏性,在精简模型结构的同时,避免冗余信息中的噪声过多的拟合到模型参数中,进而采用粒子群算法优化模型部分参数.将此方法分别应用于Mackey- Glasss混沌模型的时间序列预测和乙烯精馏塔塔釜乙烯浓度预测,应用结果表明该方法较最小二乘支持向量机、稀疏最小二乘支持向量机以及混合核最小二乘支持向量机具有更好的泛化效果和预报精度,兆示出其良好的应用潜力.

Soft Sensor Modeling and Application Based on Mixed Kernel Function and Sparse LS-SVM

Aiming at the problem that the Least Squares Support Vector Machine(LS-SVM)based on single kernel function has some limitations on performance and lacks sparseness,this paper provides a new soft-sensor modeling method based on mixed kernel function and Sparse LS-SVM.The mixed kernel is constituted by Polynomial kernel and RBF kernel.It can be both the global fitting ability and local fitting ability of least squares support vector machine.An algorithm called vector base learning has been used to improve the sparseness of the LS-SVM.At the same time simplifying the model structure,it could avoid the noise in the redundant information fitting to the model parameters.Part of the model parameters are selected by particle swarm optimization(PSO)algorithm.This method is applied to predict the time sequence of Mackey-Glasss chaos model and the ethylene consistence at the bottom of the ethylene rectifying column.The results indicate that the generalization performance and forecast accuracy of this method is better than LS-SVM,Sparse LS-SVM and mix kernel LS-SVM,and show its good potential for application.