基于改进UKF的滚动时域估计方法及其在PVC聚合过程中的应用

An improved UKF-based moving horizon estimation strategy and its application for a poly(vinyl chloride) polymerization process

聚合过程具有高度非线性和时变性等特点,参数在线估计有助于聚合过程控制性能和优化效果的改善。滚动时域估计(MHE)方法是一种用于聚合过程参数和状态估计的有效方法。本文提出了一种基于改进无迹卡尔曼滤波(UKF)的滚动时域估计方法,用于氯乙烯聚合过程机理模型时变参数的估计。滚动时域估计方法的关键问题之一是抵达成本(Arroval Cost)的近似估算,文中采用2种采样策略来实现抵达成本的自适应计算和更新。将提出的方法应用于氯乙烯聚合过程传热系数的在线估计,并与传统的滚动时域估计方法相比较,体现了该方法的有效性。

Online parameter estimation is helpful to enhance the control performance and operation optimization of polymerization processes, which are characterized by its highly nonlinearity and time-varying parameters. A moving horizon estimation （MHE） strategy is a more efficient manner for parameter and state estimation, especially for polymerization processes. In this paper, an MHE strategy based on improved unscented Kalman filter （UKF） is proposed to estimate time varying parameters in a mechanism model of a poly （vinyl chloride） polymerization process. Approximation of the arrival cost in MHE formulation is a critical issue in this domain. Two sampling strategies is used and switched adaptively to eompute accurately and update the arrival cost parameters. The proposed method is illustrated and compared with the traditional MHE approach for the online parameter estimation of heat transfer coefficient for the poly （vinyl chloride） polymerization process.