异类多模型动态融合软测量方法研究与应用

Research and Application of Soft Sensor Method Based on Heterogeneous Multi-model Dynamic Fusion

提出一种基于动态Gauss-Markov估计的多模型融合软测量建模方法。分别使用静态模型RBF网络、动态模型OELM和OLS-SVM进行建模,再用动态Gauss-Markov估计进行融合。该方法的精度要高于任何一个子模型,且能够跟踪时变系统的动态特性。将此方法应用于乙烯精馏塔塔釜乙烯浓度预测,结果表明该方法比其它方法具有更好的泛化效果和预报精度,显示出其良好的应用潜力。

A soft sensor modeling method based on multi-modeling dynamic Gauss-Markov estimation fusion was proposed.The static model RBF network,the dynamic model OELM and OLS-SVM modeling were used,and then the estimated values were fused by dynamic Gauss-Markov estimation.The accuracy of this method was higher than any sub-model.It was able to track the dynamic characteristics of time-varying systems.This method is applied to predict the ethylene consistence at the bottom of the ethylene rectifying column.The results indicate that the generalization performance and forecast accuracy of this method is better than the other methods,and has good potential for application.