摘要：AbstractThis paper develops approaches to the hands-off control problem subject to performance constraints for discrete-time linear systems. The approaches minimize thel1-norm of the control input to acquire the hands-off property, while satisfying the performance constraints that are given in terms of the quadratic cost of states and inputs with respect to the optimal solution to the finite-horizon linear quadratic regulator problem. We consider three kinds of the input and state matrices for the system; 1) known, 2) uncertain but contained in a known discrete set, and 3) uncertain but contained in a known polytopic uncertainty set. For the first two cases, we show that each problem is formulated as anl1optimization that is expressed as a second-order cone programming. We also show that the last case leads to a second-order cone programming after relaxations. A numerical example is included to illustrate the validity of the proposed approach.
关键词：KeywordsRobust controlUncertaintyLinear optimal controlConvex programmingDiscrete-time systems
