Loading...

Development of a Smart Learning-Based Distributed Load Alleviation System for Future Generation of Aeroelastic Wings

Khodabakhsh, Amir Hossein | 2024

0 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 57170 (45)
  4. University: Sharif University of Technology
  5. Department: Aerospace Engineering
  6. Advisor(s): Pourtakdoust, Hossein
  7. Abstract:
  8. In the pursuit of designing next-generation aircraft, several key challenges such as emission reduction, environmental compatibility, and energy consumption optimization are paramount. These challenges are addressed simultaneously in a multi-objective optimal design process, often leading to the emergence of innovative architectures. These architectures aim to enhance aerodynamic efficiency, extend range, minimize fuel consumption, and consequently, reduce aircraft emissions. One of the pivotal factors in augmenting the overall efficiency of an aircraft is the reduction in weight and thinning of the wings. This reduction in mass increases the slenderness ratio and leads to structures with significant flexibility. Consequently, their aeroelastic behavior intensifies in response to disturbances. Hence, a critical challenge for next-generation aircraft is the development of intelligent systems to mitigate load and control wing vibrations. In this context, this research presents the development of a ‘smart wing’. The term ‘smart’ refers to the wing’s ability to understand its behavior’s probability distribution in response to random disturbance loads and take appropriate actions to reduce the load while maintaining its structural health and integrity. This understanding is achieved through the calculation and propagation of the probability distribution function over time. The estimation and distribution of these functions for wing states, coupled with the integrated design of the actuator in the intelligent wing airfoil, lead to a load reduction system. This system actively eliminates wing structure vibrations during the aircraft’s lifespan while considering uncertainty and control effort. A novel approach employed in this research involves using physics-informed deep network models and learning algorithms to calculate the probabilistic model of the smart wing and determine the controller input in the system. A salient feature of this approach is its adaptability; the trained network model can be adapted to the real system. The application of machine learning algorithms and this proposed method can potentially maintain wing surface vibrations at an acceptable threshold by considering potential losses, changes in structural characteristics, and disturbances caused by flight over time. This could also lead to the formulation of guidelines for periodic maintenance cycles
  9. Keywords:
  10. Aeroservoelasticity ; Vibration Control ; Stochastic Dynamics ; Deep Learning ; Load Mitigation ; Fuel Consumption Reduction

 Digital Object List

 Bookmark

No TOC