Abstract
An approach to identification of dynamical model with real-time experiment design is proposed in this paper. The approach consists of parameter identification loop and receding horizon optimization loop for real-time design of experiment (DoE). D-optimality is chosen as cost function for the DoE and C/GMRES algorithm is exploited to solve the receding horizon D-optimization problem, which enables us to combine the DoE algorithm with the parameter identification loop. Finally, application result in identification of transient model of a combustion engine is demonstrated.
