针对石化行业中软测量建模样本的特性,提出一种基于在线聚类和v-支持向量回归机(vSVR)的多模型软测量建模方法。在vSVR建模过程中,通过在线聚类算法改善了vSVR模型参数选择算法的稳定性,并用vSVR参数的先验知识和KKT条件实现模型参数的快速寻优,提高了模型的学习效率和精度。该建模方法在加氢裂化分馏塔装置的轻石脑油终馏点在线预测系统中取得了良好的效果。
In order to use the properties of samples, a soft-sensing method with multiple models based on an online clustering arithmetic and v-support vector regression (vSVR) was presented. The parameter selection of vSVR is improved faster and robust by a new cross validation method using the online clustering arithmetic and parameter' s prior knowledge. The proposed soft-sensing method was used to predict the light naphtha end point in hydrocracker fractionators. Practical applications indicated the proposed method was useful for the online prediction of quality specifications in chemical processes.