Seminari - "Generalized Parameter Estimation-based Observers (GPEBO): Comparison with Other Methods and a Unifying Framework". Prof. Romeo Ortega

Lunes, 13/06/2022, 10h - 13h

In the first part of the talk we present a new approach to state observation whose main idea is to translate the state estimation problem into one of estimation of constant, unknown parameters. The class of systems for which GPEBO is applicable is identified via two assumptions related to the transformability of the system into a suitable cascaded form and our ability to estimate the unknown parameters. The first condition involves the solvability of a partial differential equation while the second one requires some excitation conditions. The estimation of the parameters is carried out using the powerful Dynamic Regressor Extension and Mixing (DREM) technique that translates the identification of a q-dimensional vector into q scalar estimation problems, whose convergence is guaranteed with the weakest possible excitation condition.

GPEBO is shown to be applicable to position estimation of a class of electromechanical systems, for the reconstruction of the state of power converters and for observation of the state of power systems equipped with PMUs. Moreover, its successful application to linear time-varying systems is guaranteed imposing only the necessary condition of observability. Extensions to the case of unknown parameters, delayed measurements and rejection of external perturbations are also discussed.

In the second part of the talk we present PEBO in a unified framework together with the— by-now classical—Kasantzis-Kravaris-Luenberger and Immersion and Invariance observers. The performance of the PEBO is compared with the one of a high gain observer in a power converter. As expected, it is shown that the performance of the latter design is significantly below par with respect to the PEBO technique.