基于卡尔曼滤波器的间歇反应预测控制

Kalman Filter Based Model Predictive Control for Batch Reactor

在间歇反应过程的温度跟踪控制问题中,为了实现全工况范围内反应器温度设定值的精确跟踪,设计了一种改进的预测控制算法。在反应自身产热对系统输出影响未知的情况下,用带有输入干扰的系统模型来建模夹套温度与反应温度之间关系,并将该干扰项增广为状态变量,然后结合线性卡尔曼滤波器估计出干扰的大小及其对系统输出的影响;结合估计出的干扰,在DMC控制算法对其加以补偿。最后在半实物的仿真环境中,验证了该控制算法在间歇温度控制中的效果,结果表明,该控制方法在一定程度上克服了反应产热对温度跟踪精度的影响,改善了控制效果。

The temperature tracking issue of batch reactor is discussed. To track the reactor temperature profile precisely in the whole working condition, a improved predictive control algorithm is designed. Under the condition of the unknown effect on the output from the heat generation, the system model with input disturbance is used to model the relationship between the temperature of the jacket and the reactor, and the disturbance vector is augmented as a system state. The linear Kalman filter is used to estimate the disturbance and its effect on the system output. With the disturbance estimated, a DMC algoritbm with compensation for the disturbance is obtained. An experiment in a hardware-in-loop simulation （HILS） environment presents a good control performance. The result shows the effect iveness on the precision of temperature tracking from the heat generation.