将基于微分-代数方程组的机理模型作为被控模型,利用联立法中的配置点法将预测控制中的滚动时域最优控制问题转化成求解非线性规划(NLP)问题.针对求解NLP问题过程中出现的由于方程维数过高而导致的求解速度过慢的现象,提出基于斐波那契数列的稀疏化策略,可以有效地减少计算量,加快求解速度.最后文中在强非线性对象连续搅拌反应釜上仿真,将该算法与标准NMPC算法和线性预测控制动态矩阵算法进行对比,体现出新算法的优越性.
This article presents an improved nonlinear model predicative control algorithm based on the rigorous model. Process plants can be described as differential-algebraic equations which are rigorous models. A collocation method of a simultaneous method is used to transform the optimal control problems into the nonlinear programming problems. Aimed at the problems of the low solution speed due to large dimensions,a sparse technology based on the Fibonacci sequence is presented to diminish the computation complexity and to accelerate the solution speed. Finally,by a simulation on the continuous stirred reactor model with strong nonlinearity we compare this algorithm with a standard NMPC algorithm and a linear predicative control algorithm of the dynamic matrix controller and the results show this algorithms superiority.