中文摘要：

预测控制可以直接解决约束以及多变量耦合问题,但其在处理上述问题时在线计算量比较大。为了减少计算量,使用了一种改进的多变量约束预测控制算法。首先引入控制增量的期望衰减因子,使预测控制着重于第一步控制增量的计算,然后使用一种基于梯度下降的迭代算法来处理约束。最后为了验证改进算法的有效性,搭建了基于MATLAB的半实物仿真平台,并将所使用的算法在上述仿真平台上实现,用于带耦合的双容水箱的实时控制。实验结果表明,所使用的算法具有很快的跟踪速率以及可以有效性的应对带约束的多变量耦合系统。实验为以后向更复杂的工业现场应用提供了一些基础,具有一定的实用价值。

英文摘要：

Predictive control can resolve the problem of constraint and multivariable coupling directly,but the calculated amount of the quadratic programming is big when it is applied in the multivariable system. So in this paper,an improved constrained predictive control is used to reduce calculated amount. Firstly,an attenuation factor of control increment is introduced,and predictive control will focus on the calculation of first control increment. Then,an iterative algorithm based on gradient descend is adopted to deal with constraint. Finally,in order to verify the effectiveness of improved approach,a semi- physical simulation system based on MATLAB is built. The proposed method is realized in the simulation system,and it is adopted to control the coupled double tank. Experimental result shows that the proposed method has high tracking speed and can control the constrained multivariable system effectively. This experiment provides some theoretical basis for further application in the complex industrial site.