Loading...

Application of Artificial Neural Networks for Identification and Modeling of Time Periodic Flapping Wing Systems

Mostafa Gharebaghi, Shabnam | 2020

407 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 53708 (45)
  4. University: Sharif University of Technology
  5. Department: Aerospace Engineering
  6. Advisor(s): Banazadeh, Afshin
  7. Abstract:
  8. For complex systems, it is difficult to obtain an accurate model of system dynamics using physical laws. Therefore, one of the most common solutions is to use system identification techniques. Flapping wings are one of the most challenging systems in the engineering world. In this study, we use a fuzzy wavelet neural network to identify time periodic flapping wing systems. In order to extract network’s input and output information, we use analytical simulation using MATLAB software. After validating the input and output information of the system with the help of the ADAMS software, the identification process of the flapping wing system is performed in longitudinal and lateral-directional mode. Identification in longitudinal mode is done with a three-input and three outputs network and identification in lateral mode is done with a network of six inputs and six outputs. A combination of a sliding mode and genetic algorithms is used to adjust the network parameters and the stability of learning algorithm can be assessed using the Lyapunov stability theorem. Finally, a comparison is made between the results of the recurrent wavelet network. The results of the designed network, and the results show that the designed network has reached a better performance with a higher convergence rate than the recurrent wavelet networks. As a result, this algorithm is suitable for real-time applications
  9. Keywords:
  10. Genetic Algorithm ; Sliding Mode ; Fuzzy Wavelet Neural Networks ; Time Periodic Flapping Wing Systems ; Dynamics Identification

 Digital Object List

 Bookmark

No TOC