[home] [Personal Program] [Help]

---

Session: Aeroelastic testing 4
Session starts: Wednesday 19 June, 16:00
Presentation starts: 17:30
Room: Room 1.2

---

Gabriel Buendia (Airbus Defence)
Beatriz Pulido (Airbus Defence and)
MIguel Torralba (Airbus Defence and Space)
Manuel Reyes (Airbus Defence and Space)
Felix Arevalo (Airbus Defence and Space)
Hector Climent (Airbus Defence and Space)

Abstract:
Artificial neural networks are known for solving complex problems and detecting nonlinear relationships between variables in a fast and accurate manner. Moreover, their storage memory optimization makes them an attractive tool for aeronautical applications such as control algorithms, component health monitoring and simulators [1]. In order to study the feasibility of applying feedforward backpropagation networks to dynamic loads problems, two applications on a military transport aircraft have been analysed: 1. Potential landing overload events assessment, where classification neural networks were used. 2. Fatigue turbulence loads prediction, where regression neural networks were used. The following methods have been explored for the proper development of these neural networks: 1. Exploration of the database inputs and outputs. This includes an initial assessment of the relevance of the input parameters, the selection of the relevant outputs to be monitored and the identification of the most dense regions in the database. 2. Neural training and hyperparameter tuning using Keras/TensorFlow 2.0. Sensitivity studies are performed to select the combination of the parameters that specify the details of the learning process which provide the best results, either for predicting a single or multiple outputs simultaneously. The definition of the cost function and metrics is of special relevance. 3. Interpolation and extrapolation capabilities assessment. Evaluations on the ability of the trained networks to predict results from non-trained inputs, which are inside or outside the limits of the variable space in which they have been trained, were performed. 4. Performance evaluation with recorded flight data. Influence of the errors coming from conservative estimations of the neural networks on the fleet operations is evaluated. The extension of the methodology to similar dynamic loads problems such as fatigue discrete gust and taxi loads, as well as any other overload events are planned to be explored in the future. References: [1] Faller, W. E. & Schreck, S. J.: Neural networks: Applications and opportunities in aeronautics. Progress in Aerospace Sciences 32.5, 433-456 (1996)