Abstract

Nowadays model-based fault detection and isolation (FDI) systems have become a crucial step towards autonomy in aerospace engineering. However few publications consider FDI applications to unmanned air vehicles (UAV) where full-autonomy is obligatory. In this paper we demonstrate a sensor fault detection and accommodation (SFDA) system, which makes use of analytical redundancy between flight parameters, on a UAV model. A Radial-Basis Function (RBF) neural network (NN) trained online with Extended Minimum Resource Allocating Network (EMRAN) algorithms is chosen for modelling purposes due to its ability to adapt well to nonlinear environments while maintaining high computational speeds. Furthermore, in an attempt to reduce false alarms (FA) and missed faults (MF) in current SFDA systems, we introduce a novel residual generator. After 47 minutes (CPU running time) of NN offline training, the SFDA scheme is able to detect additive and constant bias sensor faults with zero FA and MF. It also shows good global approximation capabilities, essential for fault accommodation, with an average pitch gyro estimation error of 0.0075 rad/s.

Highlights

  • FAULT tolerant flight control systems (FTFCS) can be found in many air-vehicles nowadays

  • In this paper we focus on sensor failures in an unmanned air vehicle (UAV)

  • This is an indication of the false alarm characteristics, and will be referred to as ‘false alarm’ (FA)

Read more

Summary

INTRODUCTION

FAULT tolerant flight control systems (FTFCS) can be found in many air-vehicles nowadays. It is apparent that this method has serious weight, power and cost implications especially for UAVs. it is apparent that this method has serious weight, power and cost implications especially for UAVs For this reason, over the past two decades analytical redundancy has become a more appealing approach for SFDA where a residual, which at its simplest form computes the difference between the model estimate and sensor measurement, is generated and a fault is declared when it exceeds a predefined threshold. Novel approaches include nonlinear online adaptive schemes where the model is continuously tuned to fit the timevarying system This is why the use of neural networks (NN) with online learning capabilities is steadily growing in the fault detection field [10]. The residual will always be nonzero due to unknown inputs (e.g. measurement noise and disturbances etc.) This can increase the risk of false alarms especially if simple threshold logic is implemented for residual evaluation. Comparisons of a conventional residual generator approach and our novel approach are carried out under different levels of sensor noise and fault classes in order to test their robustness and sensitivity respectively

UAV MODEL
NN Inputs
Network Size Reduction
RBF NN Structure and Training
Residual Padding
RESIDUAL GENERATION
FAULT SCENARIOS
SFDA OUTLINE
RESULTS
VIII. SUMMARY OF RESULTS
CONCLUSION

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call

Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.