A strong tracking filter based multiple model approach for gas turbine fault diagnosis

摘要

In this paper, a nonlinear Fault detection and isolation (FDI) based on an improved Multiple model (MM) approach was proposed for the gas turbine engine. A bank of Strong tracking extended Kalman filters (STEKFs) was designed that enables robustness to model uncertainty and overcomes the shortcoming of the MM approach. The Jacobian matrix used in the filters was deduced by using the non-equilibrium analytic linearization method to improve the traditional method. Hierarchical fault detection and isolation architecture based on evaluating the maximum probability criteria were developed for both single and multiple faults. In addition, a nonlinear mode set automatic generation method that enables automatic generation of the modes of each level in the hierarchical architecture was also presented. Fault detection and isolation of a two-shaft marine gas turbine was studied in a simulation environment using the proposed STEKF-based MM approach and compared with the results of the traditional Extended Kalman filter (EKF) based MM approach. The results showed that the proposed approach not only has the advantages of the EKF-based MM approach but also robustness to the model uncertainty and overcomes the shortcomings of the MM approach.