一类混合迭代算法在约束预测控制中的应用

为了改进多变量约束预测控制的滚动优化算法,对路径跟踪法和粒子群算法进行了理论研究,提出了一种将路径跟踪法和粒子群算法相结合的混合迭代算法,并用该混合算法对最具代表性的动态矩阵控制进行了滚动优化.仿真结果表明:对由于参数选择引起矩阵奇异时,路径跟踪法无法求出最优解,而基于混合迭代算法的优化算法可以得到最优解.采用路径跟踪法和粒子群算法相结合的混合迭代算法对多变量约束预测控制进行滚动优化,保证了优化算法的快速性和可行性.

A class of mixed iterative arithmetic for constrained predictive control

To improve rolling horizon optimization of multivariable constrained predictive control, path-following method and particle swarm optimization have been studied and a class of mixed iterative arithmetic composed of both is proposed.This method has been applied to dynamic matrix control,which is representative in predictive control.The simulation results indicate that optimum solution can be obtained by this mixed iterative arithmetic whereas it cannot be achieved in path-following method,when the matrix is singular.So this class of mixed iterative arithmetic is fast and feasible in solving multivariable constrained predictive control.