摘要
提出一种基于核主元分析(KPCA)和最小二乘支持向量机(LSSVM)的软测量建模方法。利用核主元分析对软测量的输入数据进行数据压缩,提取非线性主元,然后用最小二乘支持向量机进行建模,降低模型复杂度,提高模型泛化能力,通过交叉验证方法对支持向量机的参数进行选择。将其应用于石油树脂粘度的软测量建模,仿真结果表明,该方法具有跟踪性能好,泛化能力强等优点。与实际生产中使用的方法相比,预测精度明显提高,是一种有效的软测量建模方法。
A soft sensor based on kernel principal component analysis(KPCA) and least square support vector machine (LSSVM) is proposed.KPCA is applied to compress data,and choose the nonlinear component.LSSVM is used to proceed regression modelling,which reduces the complexity of calculation and improves the generalization ability.Cross validation method is used to select proper parameters of LSSVM method.Soft sensor is applied to predict viscosity of petroleum resin.Results show that this method features good approximation and good generalization ability.Compared with the method used in the factory now,the precision of prediction is improved.It is proved to be an efficient modelling method.
出处
《控制工程》
CSCD
北大核心
2009年第S1期176-179,共4页
Control Engineering of China
