最小二乘支持向量机(LS-SVM)是支持向量机(SVM)的一种扩展,其算法简练,计算速度快;利用LS-SVM进行特征提取,可以有效地降低输入样本维数,缩减模型的运算时间,同时LS-SVM又具有优越的非线性回归能力;为实现氧化铝高压溶出过程中苛性比值在线测量,建立了一种基于LS-SVM的软测量模型,并将此模型应用于实际生产;工业数据的仿真结果表明该模型具有较高的预测精度和范化能力,能满足在线检测、实时控制的要求。
The Least Squares Support Vector Machines(LS-SVM),a branch of Support Vector Machines(SVM),offers easier algorithm and better computability.Feature extraction by LS-SVM,can reduce the dimension of input samples and decrease the computing time of model.And also LS-SVM has ascendant capability of regression.To measure the ratio of soda to aluminate online,a soft sensing model based on LS-SVM is proposed and applied in the process of high pressure digestion of Alumina.The simulation result shows that the LS-SVM model is more precise and stronger,also it can satify the requirement of real time control.