A Metaheuristic Optimization Using Explosion Method On A Hybrid Pd2-Lqr Quadcopter Controller
Loading...
Date
2021-09-01
Authors
Raihan, Mohamad Norherman Shauqie Mohamed
Journal Title
Journal ISSN
Volume Title
Publisher
Universiti Sains Malaysia
Abstract
The popularity of the rotorcraft type UAV, the quadrotor, has grown rapidly in
recent years due to its advantages and capability to perform various applications such
as environment monitoring, surveillance, and inspection. However, the quadrotor’s
dynamics are highly nonlinear and underactuated since it has 6 DOF that need to be
controlled by only 4 actuators. Besides, it is also crucial that the controller’s gains are
tuned appropriately since it can affect the quadrotor’s performance. This study aims to
develop an effective optimal control technique to control and stabilize the quadrotor's
altitude and attitude motion. A simulation-based experiment in MATLAB/Simulink
environment was conducted to test and verify the proposed algorithm and controller
performance. The mathematical model of the quadrotor was derived based on the
Newton-Euler approach and linearized using a small angle approximation. In this
study, a Hybrid PD2-LQR controller was proposed for quadrotor control and
stabilization. Conventionally, the controller’s gains were tuned using the trial-anderror
method. The problem with this method was that it very time-consuming, and the
control designer could never tell which gains are the optimal solution for the controller.
Therefore, an optimization algorithm based on the explosion method called REA was
proposed and implemented on the proposed Hybrid PD2-LQR control structure. A
comparative study with 8 well-known algorithms, PSO, ABC, GA, DE, MVO, MFO,
FA, and STOA, was performed to evaluate the performance of the proposed algorithm.
Similarly, the proposed controller was evaluated by a comparative study with 6
conventional controllers, PD, PID, LQR, Hybrid P-LQR, Hybrid PD-LQR, and Hybrid
PD2-LQR. The findings show that the REA could perform well in exploiting the global
optimum and exploring the search space. The convergence speed of the REA was also
faster than other algorithms. Similarly, for the controller, the findings show that the
REA-based Hybrid PD2-LQR controller has a faster rise time with a shorter settling
time than the conventional controllers, while there was no overshoot and steady-state
error produced. On average, the rise time, settling time, overshoot, steady-state error
and RMSE was improved by 95%, 95.3%, 100%, 100%, and 43.5% respectively for
roll and pitch motion, while 96.5%, 96.5%, 100%, 97.2%, and 47.3% respectively for
yaw motion. For altitude motion, the rise time, settling time, overshoot, and steadystate
error were improved by 84.5%, 85.5%, 100%, and 100%, respectively. The
RMSE for altitude motion was not improved but still could be accepted since the
difference with the conventional controllers was not too much. Therefore, based on
these findings, it could be concluded that the proposed REA-based Hybrid PD2-LQR
controller was the best among the tested controller and suited for controlling and
stabilizing the quadrotor’s altitude and attitude motion since it could significantly
improve the performance of the quadrotor’s response.