参考轨迹在线优化的区间预测控制

Zone Model Predictive Control With Online Optimization of Reference Trajectory

实际生产过程控制中,特别是多变量耦合的复杂过程中,对某些被控变量其实并没有苛刻的给定值要求,只是满足在给定区间内即可,从而保证整个生产过程控制的平稳。针对这一问题,提出一种基于模型预测控制理论的区间控制算法。该算法对预测控制的优化性能指标进行改进,将性能优化指标中的参考轨迹作为约束变量进行在线优化处理;参考轨迹自动跟踪被控变量输出,使预测控制针对输出在不同的区域范围采取不同的控制强度,从而在实现区间控制的前提下满足平稳控制要求。最后以Shell公司的典型重油分馏塔控制问题为例进行仿真,验证算法的有效性和可行性。

In practical industrial control process, especially in the complex process of multi-variable and coupling performance, some controlled variables only need to satisfy the requirement of a zone control goal, which keeps the whole process stable, instead of a rigor ous set point. To solve the problem, a new algorithm of zone control is proposed in this paper, which is based on model predictive con trol. This algorithm considers the reference trajectory as constraint variable, which should be solved optimal on line. And the reference trajectory tail after the output of controlled variables automatically so that MPC can take various control to different regions of output. In this way, it realizes the stability as possible as zone control of production process. An illustrated example of typical control problem of a heavy oil fractionator proposed by Royal Dutch/Shell Group is simulated in this paper to verify the effectiveness and feasibility of the proposed algorithm.