ML Model for On-Board Helicopter Weight Estimation Developed

Accurate weight estimation is crucial for helicopter safety and performance, especially during takeoff. This research focuses on developing and implementing a novel supervised Machine Learning model to estimate helicopter weight in real-time. The model was trained using vast datasets collected from Airbus's global in-service fleet, ensuring its relevance and robustness for real-world applications. A significant aspect of this study is the detailed learning assurance process, which adheres to the European Union Aviation Safety Agency (EASA) concept paper for machine learning applications and the ongoing Eurocae ED-324 standard. This rigorous process involved defining specific Machine Learning Requirements, creating a comprehensive Machine Learning Model Description, and implementing a Long Short-Term Memory (LSTM) recurrent neural network. The researchers successfully verified these requirements against the implementation, demonstrating that the developed ML model and its deployment on legacy avionics computers are suitable for airborne targets. This paves the way for integrating such critical functions as on-board alerting systems, enhancing flight safety and operational efficiency.