摘要
针对单核函数最小二乘支持向量机容易陷入局部最优值,提出双核函数最小二乘支持向量机软测量建模算法。该方法利用Sigmoid核函数与RBF核函数线性加权构成双核函数,去除部分较小支持向量样本的方法,改善最小二乘支持向量机的稀疏特性,提高模型运算速度,用k-交叉验证法进行部分参数的优化,最后将此方法用于建立汽油干点软测量模型,并与标准支持向量机、单核最小二乘支持向量机比较,结果表明双核函数最小二乘支持向量机软测量模型具有更高的计算精度和更好的推广能力。
According to the function of mononuclear least squares support vector machine is easy to fall into local optimal value, the dual-core kernel function least squares support vector machine model method was proposed. This method uses Sigmoid kernel function and RBF kernel function which was linear weighted composing dual-core function, Removing the part of the smaller support vector sample method for improving the least squares support vector machine sparse characteristic, improving model operation speed, cross validation method was used to optimize the part of parameters. Finally the method was used to establish the gasoline dry point soft measurement model, compared with the standard support vector machine and the function of mononuclear least squares support vector machine. The results show that dual-core function least squares support vector machine model has higher calculation accuracy and better generalization ability.
出处
《石油化工高等学校学报》
CAS
2013年第3期74-78,共5页
Journal of Petrochemical Universities
基金
国家863计划项目资助(2007AA04Z162)
国家自然科学基金(61203021)资助
辽宁省科技攻关项目(2011216011)
辽宁省高校创新团队支持计划项目(2009T062
LT2010058)资助
关键词
软测量
最小二乘支持向量机
双核函数
干点
Soft sensing
The least squares support vector machine
Dual core function
Dry point
