An Adaptive Threshold Algorithm Based on Satisfactory Filtering Principle

Abstract

In this paper, the problem of designing a adaptive threshold for Fault diagnostic in Full Authority Digital Engine Control system that are subject to environmental disturbances and various input is investigated. According to aero-engine small deviation linearization model, state matrix perturbation model with "soft boundary" pseudo normal distribution is built. The proposed algorithm bases on satisfaction filtering principle. Linear matrix inequality (LMI) method is applied to get upper bound of residual. With the residual distribution probability model and the target false alarm rate, the optimal threshold is determined. The proposed algorithm is then applied to sensor fault detection on a turbofan aero-engine. Simulation results demonstrate and validate the performance and capability of the adaptive threshold algorithm.