实时优化与分布式控制集成算法

Integration of Real-time Optimization and Distributed Predictive Control

预测控制作为一种以预测模型为基础的先进控制算法,分布式预测控制广泛应用于复杂高维的复杂大系统控制中;提出了针对大规模过程的实时优化与分布式预测控制集成算法,包含稳态目标计算层和动态分布式控制层;在稳态目标计算层根据当前系统的运行状况进行集中优化,在系统运行的每一时刻计算出相对的全局最优值,并将其传递到下层进行控制;在动态控制层将复杂的大系统分为若干个相对独立的子系统,并且充分考虑各个子系统之间的关联和耦合,用分布式预测控制算法对上层计算得到的相对的全局最优值进行跟踪,提高系统在动态情况下的控制性能;仿真应用表明,此方法的优点在于保证全局最优的同时,降低了计算的复杂度,并且实现了经济目标。

Predictive control has been widely applied as an advanced process control methodology in complex industrial processes control. Distributed predictive control was mainly proposed for complex and high dimensional large scale system. In this paper, an integrated algo- rithm of real time optimization and distributed predictive control is put forward for large scale processes, which includes one layer of steady state objective calculation and another layer of dynamic control. Centralized optimization is implemented in the layer of steady state objective calculation, in which steady optimal value is calculated. Then steady optimal value is conveyed to the layer of dynamic control. In the layer of dynamic control, complex and large scale system is divided into multiple subsystems fully considering the association and coupling. For im proving dynamic performance of system, distributed predictive control is used to track the set point from the layer of steady state objective calculation. The simulation shows the advantages of this method in global optimizaion, reducing complexity of calculation and achieving the economic goals.