[Paper Review] Active Fault Tolerant Flight Control System Design - A UAV Case Study

Rudaba Khan, Paul Williams+3 more|arXiv (Cornell University)|Oct 11, 2016

Fault Detection and Control Systems6 references3 citations

TL;DR

This paper proposes an active fault tolerant control (FTC) system for fixed-wing UAVs using analytical redundancy via nonlinear model predictive control (NMPC) and an unscented Kalman filter (UKF) for fault detection and identification (FDI). The system identifies engine failures within seconds and redistributes control authority to healthy actuators, maintaining safe flight without hardware redundancy.

ABSTRACT

Fault tolerance is achieved through multiply redundant hardware systems in large civil aircraft. This means of achieving fault tolerance is infeasible for small compact unmanned aerial vehicles. In this paper we apply a fault tolerant control system which exploits analytical redundancy rather than hardware redundancy to an actual UAV model currently in operation via model-in-the-loop simulation. The fault tolerant control system comprises a nonlinear model predictive controller integrated with an unscented Kalman filter for fault detection and identification. The results show that our fault tolerant control system design is able to identify engine failure within seconds of fault occurrence and distribute control authority to the healthy actuators to maintain safe flight.

Motivation & Objective

• Address the impracticality of hardware redundancy in compact, small UAVs due to size, weight, and cost constraints.
• Develop a fault-tolerant flight control system that leverages analytical redundancy instead of physical redundancy.
• Design a reconfigurable control system capable of detecting and responding to engine failures in real time.
• Validate the system using a real UAV model via model-in-the-loop simulation under realistic flight conditions.
• Demonstrate the feasibility of active FTC for fixed-wing UAVs in civil airspace operations.

Proposed method

• Integrate a nonlinear model predictive controller (NMPC) with pseudospectral numerical techniques to optimize control inputs based on real-time system dynamics.
• Use an unscented Kalman filter (UKF) with defined process noise and measurement noise covariance matrices to estimate system states and detect faults.
• Implement fault detection logic based on residual analysis and thresholding, with fault flag activation upon deviation beyond 2σ uncertainty bounds.
• Define control constraints including minimum thrust at 0N and variable upper bounds based on UKF-estimated thrust with 2σ uncertainty margin.
• Utilize the Dryden Wind Model in MATLAB to simulate realistic wind effects during flight trajectory tracking.
• Apply a reference trajectory for path following and evaluate system performance under no-fault and two engine failure scenarios (50% and 70% power loss).

Experimental results

Research questions

• RQ1 Can analytical redundancy be effectively exploited to achieve fault tolerance in small UAVs without hardware redundancy?
• RQ2 How quickly can the proposed FTC system detect and identify engine failures during flight?
• RQ3 To what extent can the NMPC-UKF system reconfigure control inputs to maintain stable flight after an engine failure?
• RQ4 How does the system perform under varying levels of thrust loss (50% and 70%) in terms of trajectory tracking and airspeed regulation?
• RQ5 Can the system prevent stall and maintain safe flight characteristics despite significant power degradation?

Key findings

• The fault detection system identified engine failures within 2 seconds of occurrence, as indicated by the fault flag switching to 1.
• For a 50% power loss, the true airspeed demand of 20 m/s was not achieved during climb but was gradually reached during straight and level flight and maintained during descent.
• With 70% power loss, the aircraft failed to meet the airspeed demand during straight and level flight but regained speed halfway through the descent phase.
• The aircraft did not reach stall speed in any scenario, indicating sustained safe flight margins.
• During power loss, the climb rate oscillated between upper and lower constraints due to increased elevator authority, indicating effective but aggressive control reconfiguration.
• In the 70% power loss case, the aircraft could not reach the target altitude and instead maintained a lower cruising altitude, later rejoining the reference trajectory during descent.