Uma nova abordagem de identificação por subespaços com imposição de modelos passivos

Abstract

The parameter estimation of systems that operate interconnected requires not only accuracy but also the preservation of properties intrinsic to the physical system, as passivity, in such a way that models that do not guarantee that property may incur unstable behavior in time-domain simulations. The main objective of this paper is the application of a 4SID algorithm to estimate passive state-space models, from frequency response data. To attain that purpose, passivity enforcement equations based on LMIs are formulated as constraints of the optimization problem to reach a globally optimal and passive solution. A weighting methodology, in the frequency domain, is also formulated as an LMI, and it is incorporated into the algorithm in order to improve the approximation between the response of the estimated model and the data. A case study using experimental data measured from a power transformer is presented to show and validate the proposed approach. The results reveal that the proposed algorithm solves passivity enforcement problems in a single step without interactions and is capable to estimate passive models with 4SID algorithms.