Neural Network based architecture for Fault Detection and Isolation in air data systems

This paper presents the design, development, integration and flight testing of a Fault Detection and Isolation architecture for an air data computer based on Artificial Neural Networks. A lot of Networks have been trained using Knowledge Discovery in Data Base Process in order to identify faults on air data measurements such as airspeed, sideslip angle, and angle of attack. The proposed methodology makes use of a huge number of flight data for training and testing in the Neural Network design. Flight data have been recorded during flight trials carried out using the experimental aircraft of the Italian Aerospace Research Centre. The proposed architecture tested on flight data gathered during an autonomous mission of an Unmanned Aerial Vehicle (UAV) shows good performance in identifying fault occurrences.