Robust analysis and design of control systems using interval arithmetic (Q1367196)

One presents, compares and tests (on five examples) some interval arithmetic methods for the analysis and design of robust control systems, affected by nonlinear parametric uncertainties and nonparametric perturbations. In the case of polynomial perturbations, the subdivision algorithm based on Bernstein polynomial expansion appears quite efficient. It is underlined that the computational complexity grows exponentially with the number of parameters, while methods exploiting simpler parameterizations (e.g. the linear one) do not have this drawback. The choice between linear and nonlinear approaches is a matter of trade-off between complexity and conservativeness. The techniques presented in the paper are able to deal with up to ten plant parameters in robust analysis problem plus three controller parameters in robust design problem for computation times not exceeding one hour on a workstation.