Loading...

Self-Tuning PID via a Hybrid Neural Structure for Attitude Control of Quadcopter

Sharifi, Iman | 2022

197 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 54868 (08)
  4. University: Sharif University of Technology
  5. Department: Mechanical Engineering
  6. Advisor(s): Alasty, Aria
  7. Abstract:
  8. Proportional-Integrator-Derivative (PID) controller is used in a wide range of industrial and experimental processes. There are a couple of offline methods for tuning PID gains. However, due to the uncertainty of model parameters and external disturbances, real systems need more robust and reliable PID controllers. A compelling example of these types of systems is Quadrotor. In this article, a self-tuning PID controller using Reinforcement Learning for attitude control of a quadrotor has been investigated. In the proposed method, an Incremental PID, which contains constant and variable gains, has been used, and only the variable ones have been tuned.In this research, the model-free Actor-Critic algorithm with an efficient Hybrid Neural Structure was used that not only been able to properly tune PID gains but also done the best as an identifier. In both tunning and identifying tasks, a neural network with two hidden layers and sigmoid activation functions has been learned using Adaptive Momentum (ADAM) Optimizer and Back-Propagation (BP) algorithm. This method is online, able to tackle disturbance, and fast in training. In addition to robustness to mass uncertainty and wind gust disturbance, results showed that the proposed method had a better performance when compared to a PID controller with constant gains
  9. Keywords:
  10. Reinforcement Learning ; Quadrotor ; Neural Network ; Hybrid Neural Network ; Actor-Critic Model ; Self-Tuning Proportional-Integral-Derivative (PID)Controller

 Digital Object List

 Bookmark

No TOC