Enhanced Neural Network Model for Aircraft Flight Dynamics Prediction from Flight Test Data- Business Aircraft Application

The ability to effectively detect data anomalies during flight test campaigns is highly dependent on the availability of a highly accurate mathematical model of an aircraft. However, identifying a model capable of accurately predicting the dynamics of an aircraft across the entire flight envelope remains challenging for engineers. This difficulty arises not only from the diversity of maneuvers performed during flight tests but also from the inherent complexity of aircraft dynamics. To address this challenge, a study was conducted at the Laboratory of Applied Research in Active Control, Avionics, and AeroServoElasticity (LARCASE) to propose an innovative methodology for identifying a model for a specific aircraft using Neural Networks. This artificial intelligence-based model was specifically designed to learn and predict the aircraft f light dynamics under a wide range of operating conditions. Compared to traditional time-series forecasting models, the proposed model aims to enhance generalization across various flight maneuvers and enable long-term forecasting. Results showed that the trained model was able to predict a set of aircraft flight parameters very well and within a predefined set of tolerances. By capturing long-term dependencies and efficiently generalizing model capabilities to any flight tests, the proposed model offers a promising solution to the problem of anomalies detection in f light testing.