Evolutionary programming for minimum uncertainty parameter estimation for linear models

In this paper linear parameter models with stochastic and set membership error descriptions are considered. A genetic algorithm is presented for the determination of a fixed size measurement set that allows one to avert the case of unidentifiability and permits to estimate the system parameters with the smallest uncertainty of the parameters. Such an approach is relevant in experimental design when an optimal sampling schedule is sought. A numerical simulation study has been carried out to evaluate the performances of the algorithm.