A DESIGN OF ROBUST NONLINEAR CONTROL SYSTEM -APPLICATION TO THE CONTROL OF TEMPERATURE DISTRIBUTION OF PACKED BED REACTOR-

通过测定以环氧乙烷为主要产物的填充床反应器(内径20.4 mm，长800 mm)的出口浓度，确定了反应动力学。该反应体系由一个产生环氧乙烷的主反应和一个产生二氧化碳的副反应组成。通过测量环氧乙烷和二氧化碳的出口浓度来确定这些反应的动力学，并在填充床反应器的物料平衡和热平衡的基础上建立了数学模型。确定的反应速率被包含在数学模型中。仿真结果表明，在反应器入口附近出现了一个热点，动态行为与实验结果一致。为了估计反应器内温度和浓度的分布，利用该数学模型，基于扩展卡尔曼滤波算法设计了状态观测器。实验表明，所设计的观测器仅需6~8次温度测量即可获得较好的效果。此外，即使数学模型包含一定的建模误差，也可以正确地估计出浓度。然后，该观测器可以应用于实际过程。

By measuring outlet concentration of the packed bed reactor (20.4 mm i.d., 800 mm long) which produced ethylene-oxide as a main product, we identified the reaction kinetics. The reaction system consists of a main reaction producing ethylene-oxide and a sub-reaction producing carbon dioxide. The kinetics of these reactions were identified by measuring the outlet concentrations of ethylene-oxide and carbon dioxide.The mathematical model was developed on the bases of the material balance and the heat balance of the packed bed reactor. The identified reaction rates were included into the mathematical model. The simulation results showed that a hot spot appeared near the inlet of the reactor and the dynamic behavior was consistent with the experimental one.In order to estimate the distribution of temperature and concentration in the reactor, a state observer was developed based on the Extended Kalman Filter algorism by use of the mathematical model.From the experiments, it was demonstrated that the designed observer performed well by using only 6 to 8 measurements of temperature. Furthermore, the concentration was estimated correctly even though the mathematical model involved some modeling error. Then this observer may be applied to the real processes.