Virtual sensing of wheel direction from redundant sensors in aircraft ground-steering systems

• Published in: CEAS aeronautical journal Vol. 13; no. 1; pp. 199 - 213
• Main Authors: Dal Borgo, Mattia, Elliott, Stephen J., Ghandchi Tehrani, Maryam, Stothers, Ian M.
• Format: Journal Article
• Language: English
• Published: Vienna Springer Vienna 01.01.2022; Springer Nature B.V
• Subjects: Aerospace Technology and Astronautics; Aircraft; Aircraft control; Algorithms; Computation; Control systems; Engineering; Faults; Fuzzy logic; Ground based control; Landing gear; Least squares; Noise measurement; Original Paper; Position measurement; Remote sensors; Safety critical; Sensors; Steering; Uncertainty; Soft-computing; Fault detection and isolation; Digital twin; Redundant sensors; Virtual sensor
• ISSN: 1869-5582; 1869-5590
• DOI: 10.1007/s13272-021-00557-z

Many safety-critical control systems use multiple redundant sensors to estimate the same controlled signal. If the sensors were to operate perfectly, only a subset of them would need to be used for the estimation. In practice, however, the sensors are subject to uncertainty, minor or major faults and their operation may be nonlinear. It is thus important to reliably estimate the controlled signal under these conditions, and also to assess the degree of confidence with which each sensor should be treated. An example of such a control system is the ground-steering control system of an aircraft nose landing gear. A virtual sensing technique is commonly employed, which estimates the steering angle using the measurements of multiple remote displacement sensors. The wheel position is then calculated as a nonlinear function of these sensor outputs. This paper describes how a digital twin of the ground-steering system, in which the effects of uncertainties and faults can be systematically analysed and studied, is used to assess the accuracy and integrity of the steering angle estimation for a number of different estimation algorithms. Two of these algorithms are based on a least-squares approach, while another is a soft-computing technique based on fuzzy logic. These methods are investigated for several scenarios where model uncertainty, measurement noise and sensor faults are included. It is shown that the soft-computing approach is more robust than the least squares based methods under these conditions.