Search

This paper deals with the development of an on-line identification method for deriving simplified dynamic models. The practical applicability of the derived algorithm has been tested on the Capstone C30 microturbine installed at the Polytechnic University of Bari, Italy. For this purpose, a second order model has been identified by adopting the sensitivity theory involving the Lyapunov function giving rise to stable solutions in the on-line environment, renewing model parameters at every changes occurring on the physical system. This characteristic can be fruitfully adopted for implementing actual models in time domain simulators, giving rise to accurate results even if some internal parameters change. In order to investigate on the effectiveness of the proposed methodology in deriving a simplified dynamic model, several simulations and experimental tests were conducted in the Prince Lab of the Polytechnic University of Bari. The good agreement between simulations and laboratory tests confirm the validity and the applicability of the proposed methodology. [ABSTRACT FROM AUTHOR]

Abstract: In this paper a novel methodology for evaluating the external equivalent model of a distributed network is presented. The proposed method is based on a non-linear parameter identification process which involves an artificial dynamic system, whose dynamics are explicitly designed to be stable by adopting the Lyapunov theory. This dynamic identification method is competitive with existing reduction methods, since it does not require system linearization or any a priori information about the system under study. Few measurements at boundary nodes are required to obtain the essential information regarding the external system. Even if these measurements are strongly corrupted by noise, the algorithm is quite insensible to it. Due to its ability to be auto-adaptive at any changes in the system operating conditions and to its great speed convergence, the identification procedure can be usefully applied for real-time applications. [Copyright &y& Elsevier]