Neural fuzzy approach to F/A-18 aircraft failure isolation and reconfiguration design

Abstract

Modern high performance aircraft systems require advanced onboard fault detection, isolation and reconfiguration (FDIR) techniques against system failures under a variety of conditions. Existing FDIR designs are subject to uncert ainty, nonlinearity, and complexity, leading to high false alarm rate. Recently , the emergence of Neural Fuzzy scheme has generated a great deal of interest in system modeling and control [5],[2],[14],[10],[4],[3]. It has demonstrated its ability in coping with the engineering problem with ambiguity (fuzziness) involved like pattern recognition and system diagnosis. In this paper, we will devise two FDIR schemes based on Fuzzy Logic concept. The first one incorporate the idea of subsethood for failure isolation and the method of MMAC(Mu1tiple Model Adaptive Control) [ll] for reconfiguration controller design. The second one applies a set of fuzzy rules to dominate online parameter adaptation and reach the goal of failure isolation. The simulation results are provided by using the F/A-18 aircraft model. The advantages and disadvantages of both schemes are briefly discussed in Conclusion section.