ASSESSMENT OF SUB-OPTIMAL SDRE CONTROL SYSTEMS PERFORMANCE BY DEFINING A SET OF RANDOM WEIGHTING MATRICES

Resumo

It is well known that the design of a huge class of non-linear control systems depends heavily
on the state-space representation of the to-be-controlled system. So, the task of choosing appropriate
control parameters for the nonlinear system’s stability and performance requirements tends to be time
consuming, tedious and relies mostly on the designer’s experience. The work presented here evaluates the
system performance using sets of random weighting matrices as input for the feedback control technique
based on the State-Dependent-Riccati-Equation (SDRE). These randomly generated sets are obtained by
using Monte Carlo sampling and UQLab, then evaluated using numerical simulations of two one degree
of freedom strongly non-linear systems and a slewing flexible structure. The results showed that these
sets serve as an interesting tool to map controllable regions of interest while allowing the designer to
visualize the effects of these weights on the system’s constraints and the performance requirements of
the systems response.