基于改进Kalman滤波算法的多模型融合建模方法

针对聚合物生产过程重要质量控制指标或状态变量的软测量问题,提出了一种基于改进Kalman滤波算法的多模型融合建模方法。将混合核函数主元分析(K₂PCA)与人工神经网络(ANN)相结合,建立一种基于K₂PCA-ANN的数据驱动模型;利用改进Kalman滤波算法实现K₂PCA-ANN模型与机理模型融合,构建一种并联结构的混合模型;协调二次滤波(线性滑动平滑)和方差更新对混合模型进行优化处理,使混合模型的估计性能尽可能地达到最优,使混合模型的预测稳定性得到有效改善。将该多模型融合建模方法应用于氯乙烯聚合过程聚合速率软测量中,应用研究结果表明:与单一的机理模型或K₂PCA-ANN数据驱动模型的预测性能相比,该建模方法建立的聚合速率模型具有更佳的预测性能。该建模方法的运用为进一步开展聚合物生产过程优化与控制等研究提供基础条件。

Multi-model fusion modeling method based on improved Kalman filtering algorithm

A multi-model fusion modeling method based on improved Kalman filter algorithm was presented for soft sensor of key quality index or state variable in the polymerization process. First, a data-driven soft-sensor modeling method was proposed by combining mixtures of kernels principal component analysis (K₂PCA) with artificial neural network (ANN). Second, a parallel hybrid model was constructed by fusing the data-driven model with a mechanism model through an improved Kalman filtering algorithm. Moreover, a linear smoothing filter and a model variance updating method were adopted for optimizing the hybrid model, which could enhance performance and improve prediction stability of the hybrid model. The application of the proposed multi-model fusion modeling method in the vinyl chloride polymerization rate prediction verified that the hybrid model was more effective in comparison with single-model cases (thermodynamic mechanism model or K₂PCA-ANN model). The proposed multi-model fusion modeling method would provide basic conditions for control and optimization of PVC polymerization process in further research.