科研之友
微信
新浪微博
Facebook
分享链接
摘要
Recent research has shown that the uncertainty intervals of state variables and measurements can be estimated by solving mathematical programming problems corresponding to the maximization or minimization of a component of these variables. However, the results are not guaranteed since the optimal problems are non-convex and the global optimal solutions cannot always be obtained. This paper presents a new scheme for guaranteed state estimation. Firstly, a conic programming is formulated to model these optimal problems by redefining variables and loosing the feasible space. Secondly, interval constraints propagation with all constraints concerned is applied to contract the solutions further to eliminate the pessimism. Comparisons are carried out, and results show that the result intervals are guaranteed and small enough for the control system with control dead zone concerned.
