Fault Detection and Estimation of Fault Redundant Airspeed using Unscented Kalman Filter

Fault Detection and Estimation of Fault Redundant Airspeed using Unscented Kalman Filter

Janakiraman, Sathya (TU Delft Aerospace Engineering)

de Visser, C.C. (mentor)
Schopferer, Simon (mentor)

Delft University of Technology

Delft University of Technology

Aerospace Engineering

2023-08-25

Having a reliable flight envelope is of paramount importance for safe flight operations. Parameters that determine the safe flight envelope are airspeed , angle of attack, angle of sideslip, Euler angle and load factor. These parameters need to be fault redundant, In order to address this challenge, an unscented Kalman estimation routine is proposed involving a novel kinematic model which incorporates the effects of the turbulence. An approach in which two unscented Kalman Filter (UKF) filters operate simultaneously, one with additive faults Augmented Fault Filter (AF) and the other with non-additive faults Sensor Noise and Bias Filter (SNBF). The reasoning behind this approach is that the AF filter has the fault estimated state as the additive fault to the state estimation, which would give an accurate fault redundant sensors data estimate. To address the issue where sensor noise and biases cannot be distinguished from faults occurring in the sensor in the case of a more realistic fault, the SNBF filter is used. The proposed approach is validated with the data generated by Citation-550 Simulator 2023. This approach requires the sensor data from pitot tube, angle of attack vanes, angle of sideslip vanes, roll angle, pitch angle and yaw angle data from the Inertial Navigation System (INS) and for the kinematic equation the Inertial Measurement Unit (IMU) data (Accelerometer / Gyroscope) are used to perform the estimation. This approach can produce a fault redundant estimate of all the 6 sensor states, and in this work airspeed fault detection is the main focus. The approach is valid within the entire flight envelope, and there is no need to design a linear parameter- varying system. Through this approach, airspeed fault detection is performed and a fault redundant airspeed is estimated. The Augmented Fault UKF is found to achieve an unbiased state estimation even in the presence of unknown disturbances.

Airspeed Fault Detection
Unscented Kalman Filter
Realistic Fault Simulation
Fault Detection
Double Model Adaptive Estimation

http://resolver.tudelft.nl/uuid:75adec7e-0ea5-4521-b17a-f315bfa1b085

2023-09-01

https://www.easa.europa.eu/en/research-projects/enhanced-fault-detection-and-diagnosis-solutions-air-data-systems Unspecified The main objective of the project is to propose and develop novel methods to detect and/or be robust to multiple consistent and possibly simultaneous air data probe failures. This will improve the EASA certification standards, and support the evaluation of new designs proposed by aircraft manufacturers.

Student theses

master thesis

© 2023 Sathya Janakiraman