Computation of Robust Controlled Invariant Sets with Fixed Complexity Using Bilinear Optimization
Linear Systems; Invariant Sets; Control Constraints; Robust Positively Invariant Sets; Bilinear Optimization.
In this work, a methodology for the computation of robust controlled invariant polyhedra of fixed complexity, based on bilinear optimization, is proposed for discrete-time linear systems, subject to constraints on the states and control inputs and to bounded disturbances. In many real-life applications, certain variables in a system must comply with certain constraints. In general, these constraints are specified by linear inequalities that define limited and closed polyhedral sets. A set is robust controlled invariant if any state trajectory starting in this set can be maintained within it through a suitable control input, in spite of the disturbances. Thus, the calculation of a controlled invariant set is an important step in solving control problems under constraints. Conventional methods for robust invariant polyhedra computation can result in high complexity sets, defined by a large number of vertices. The methodology proposed in this work has bilinear invariance conditions and polyhedra represented by vertices, whose quantity is fixed in advance. The aim is also to maximize the volume of the robust controlled invariant polyhedra. Through numerical examples, the methodology is able to compute polyhedra with larger volumes than those obtained by recent methods which also seek for reduced complexity sets. In addition, a methodology is numerically efficient, applicable to larger systems than those treated by the methods available in the literature.