A novel distributed extended Kalman filter for aircraft engine gas-path health estimation with sensor fusion uncertainty

Abstract

This paper is concerned with state estimation approach to track aircraft engine gas-path health condition in an advanced distributed architecture. The sensor measurements are divided into several subsets by installation position along gas path, and they are integrated to estimate engine health state changes with sensor fusion uncertainty. The uncertain sensor fusion is characterized by time delay and packet dropout in the fusion behavior of sensor measurements, and the delay steps occur randomly. A novel distributed extended Kalman filter with the data buffer bank (DEKF) is developed, and self-tuning buffer strategy of recursive fusion estimation is combined to the DEKF to form the self-tuning DEKF (SDEKF) algorithm for improving state estimation performance. The lengths of data buffer bank related to the local filters of SDEKF are different, and they are independently adaptive to the information loss level and local estimation accuracy. Local states are calculated using the measurements collected at the latest steps in self-tuning buffer banks, and then sent to master filter to yield global state and covariance by fusion estimation. The contribution of this study is to propose a novel EKF algorithm for state estimation in the distributed framework with sensor fusion uncertainty, and it achieves better trade-off between the estimation accuracy and computational efforts. The systematical comparisons of basic EKF, constant buffer DEKF and SDEKF algorithms are carried out for aircraft engine gas-path health estimation with sensor fusion uncertainty. The simulation results show the superiority of the SDEKF, and it confirms our viewpoints in this paper.