中文摘要：

针对传统多模型软测量方法在面对复杂、多变工况时缺少在线更新机制、更新时输出精度降低等问题,提出了一种基于即时学习算法（JIT）的多模型在线软测量方法（MOSVR）。离线阶段首先采用模糊C均值聚类（FCM）对训练数据进行聚类,接着采用SVR建立初始模型集。在线部分以多模型输出作为主要输出,当出现新工况时,通过在线模型更新策略（OSMU）将输出模式切换为JIT,同时多模型集进行在线更新。该方法不仅拥有多模型输出的快速性、精确性,而且在模型更新时通过JIT模式还能保证输出的连续性、稳定性、精确性。最后将该软测量方法进行了数值仿真并运用到乙烷浓度软测量中,验证了该方法的有效性。

英文摘要：

In case of complex and changeable working conditions, traditional multi-model soft-sensor techniques lacked an online update mechanism and decreased accuracy upon updating. A new soft-sensor method based on just-in-time algorithm（JIT） and multi-model online support regression（MOSVR） was proposed. In offline phase, fuzzy C-mean clustering（FCM） was employed to classify training data and SVR was used to build initial model set. In online phase, main output was multi-model SVR works, which would be switched to JIT model by online strategy of model updating（OSMU） and the current model set was updated online simultaneously when new working condition was encountered. The new method not only possessed rapidity and accuracy of multi-model outputs, but also guaranteed continuity, stability and accuracy of JIT outputs at model updating. Method effectiveness was demonstrated by numerical simulation and application in soft-sensor measurement of ethane concentration in ethylene production.