概率约束预测控制的凸优化描述研究

his paper develops a model predictive control strategy for stochastic linearsystems with both multiplicative and additive uncertainty. The random natureof the system is characterized by a sequence of polytopes, in which thestate-transition matrix evolves with certain probability. Design of the polytopesrelies on random samples of the transition matrix, and computation of thepolytopes is efficient through a convex optimization. On the basis of thepolytopic description, linear probabilistic constraints can be transformedinto deterministic ones and be solved in low computation burden. The proposed MPCalgorithm ensures probabilistic constraints and closed-loop stability. The resultsare illustrated by a numerical example.