Abstract: In this paper, the problem of robust model predictive control (MPC) for discrete-time linear systems in linear fractional transformation form with structured uncertainty and norm-bounded disturbance is investigated. The problem of minimization of the cost function for MPC design is converted to minimization of the worst case of the cost function. Then, this problem is reduced to minimization of an upper bound of the cost function subject to a terminal inequality satisfying the l2-norm of the closed loop system. The characteristic of the linear fractional transformation system is taken into account, and by using some mathematical tools, the robust predictive controller design problem is turned into a linear matrix inequality minimization problem. Afterwards, a formulation which includes an integrator to improve the performance of the proposed robust model predictive controller in steady state condition is studied. The validity of the approaches is illustrated through a robust control benchmark problem.
Abstract: This paper proposes a new parameter identification
method based on Linear Fractional Transformation (LFT). It is
assumed that the target linear system includes unknown parameters.
The parameter deviations are separated from a nominal system via
LFT, and identified by organizing I/O signals around the separated
deviations of the real system. The purpose of this paper is to apply LFT
to simultaneously identify the parameter deviations in systems with
fewer outputs than unknown parameters. As a fundamental example,
this method is implemented to one degree of freedom vibratory system.
Via LFT, all physical parameters were simultaneously identified in this
system. Then, numerical simulations were conducted for this system to
verify the results. This study shows that all the physical parameters of a
system with fewer outputs than unknown parameters can be effectively
identified simultaneously using LFT.