Physics-guided Neural Network (PNN) feedforward control of a convertible delta-wing aircraft

TN France; Nouvelle-Aquitaine; Sur place; EUR 30 000 - 50 000; Plein temps

Résumé du poste

An innovative internship opportunity awaits you at a forward-thinking company focused on developing tilt-wing delta-wing convertible aircraft. This role emphasizes the design of feedforward control systems, leveraging advanced techniques such as physics-guided neural networks. You will engage in exciting tasks, including studying existing aircraft models and applying cutting-edge methodologies in MATLAB and Python. Join a team that is pushing the boundaries of aerospace technology, where your contributions will play a crucial role in enhancing flight efficiency and safety. If you have a passion for control systems and a desire to innovate, this internship is perfect for you.

Qualifications

• Solid understanding of control systems and experience with controller design.
• Proficiency in MATLAB/Simulink and Python is highly desired.

Responsabilités

• Study existing delta-wing aircraft model and flatness analysis.
• Understand PNN feedforward control principles and apply them.

Connaissances

Control Systems

Controller Design Methodologies

MATLAB/Simulink

Python

Nonlinear System Analysis

Neural Network Design

Formation

Engineering Degree

Master's in Control Systems

Outils

MATLAB

Simulink

PyTorch

TensorFlow

Description du poste

Job Description

Supervisors: Tudor-Bogdan Airimitoaie (University of Bordeaux), Mircea Lazar (Eindhoven University of Technology).

Convertible aircraft combine the advantages of classical airplanes (wing design to reduce fuel consumption and allow for long-distance flights) with those of multicopters (hover, maneuverability at low speed, take-off and landing without the need for a runway). There are two main convertible aircraft designs:

• tilt-rotor, where only the rotors tilt, while the aircraft body and the wings are always horizontal;
• tilt-wing, where the wings and the rotors tilt together, while the body is fixed.

We are interested in a tilt-wing delta-wing convertible aircraft capable of VTOL and hover. Its horizontal flight is energy-efficient, leveraging wind energy for flight.

From a control system perspective, the focus is on the transition phase between VTOL/hover and horizontal flight to ensure safety and minimal energy consumption.

We propose an architecture combining feedback and feedforward controllers. The internship's main goal is to design the feedforward component. Previous work indicates that the aircraft is flat in hover and horizontal flight under certain assumptions, but the flatness analyses differ between phases. These differences and assumptions reduce the effectiveness of flatness-based feedforward control during transitions. Inspired by recent advances in physics-guided neural networks (PNN) for motion control, we aim to enhance flatness-based control with neural networks to improve performance.

The internship tasks include:

1. Studying the existing delta-wing aircraft model, flatness analysis, and MATLAB/Simulink code (from previous internship).
2. Understanding PNN feedforward control principles and applying them to the delta-wing aircraft.
3. If needed, transferring MATLAB code to Python to utilize neural network design and training tools such as PyTorch or TensorFlow.

Profile: A solid understanding of control systems, experience with controller design methodologies (e.g., PID, H-infinity), proficiency in MATLAB/Simulink and Python are highly desired. Knowledge of nonlinear system analysis, flat dynamic systems, and neural network design and training is a plus.