The paper focuses on active fault diagnosis (AFD) of large scale systems. The multiple model framework is considered and two architectures are treated: the decentralized and the distributed one. An essential part of the AFD algorithm is state estimation, which must be supplemented with a mechanism to achieve feasible implementation in the multiple model framework. In the paper, the generalized pseudo Bayes and interacting multiple model estimation algorithms are considered. They are reformulated for a given model of a large scale system. Performance of both AFD architectures is analyzed for different combinations of multiple model estimation algorithms using a numerical example.
2
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
Target tracking using bistatic bearings-only measurements has obtained distinct interest recently. It is a nonlinear problem that traditional Kalman filter (KF) can not be applied directly. In this paper, the triangular ranging formula has been derived first for bistatic bearings-only tracking. The ranging error is then proved to be Gaussian noises, which enable the traditional KF applicable. The recently developed unscented Kalman filter (UKF) is also applied to the nonlinear measuring equation directly. To further improve the tracking accuracy especially in case of maneuvering target tracking, interactive multiple model (IMM) is adopted. Simulation results for both constant velocity moving target and maneuvering target are included to compare the performance of the aforementioned methods. The triangular ranging method, triangular-ranging-based Kalman filtering (TRKF), UKF, TR-IMMKF, and IMM-UKF are compared extensively using the criterion of root of the mean squared error (RMSE) and computational burden, as well as the robustness.
PL
W artykule zaproponowano formułę o zasięgu trójkątnym w zastosowaniu do bistatycznego wyznaczania namiaru (pelengu). W rozwiązaniu wykorzystano m. In. filtr Kalman’a do pomiaru wielkości nieliniowych. Wyniki badań symulacyjnych, dla obiektów w ruchu jednostajnym lub zmiennym, pozwalają na porównanie działania metod. Porównano także metody TRKF, UKF, TR-IMMKF, IMM-UKF, pod względem błędów średniokwadratowych, odporności, złożoności obliczeń.
JavaScript jest wyłączony w Twojej przeglądarce internetowej. Włącz go, a następnie odśwież stronę, aby móc w pełni z niej korzystać.