Hybrid controller with neural network PID/FOPID operations for two-link rigid robot manipulator based on the zebra...

Area of Science:

• Robotics
• Control Systems Engineering
• Artificial Intelligence

Background:

• Robotic manipulators face challenges from external disturbances, parameter uncertainties, and complex nonlinear dynamics.
• Controlling multi-input, multi-output (MIMO) systems like robotic manipulators requires advanced control strategies.
• Existing controllers may struggle with the coupled and nonlinear nature of robotic systems.

Purpose of the Study:

• To propose and evaluate six novel control structures for a 2-link rigid robot manipulator (2-LRRM) focusing on trajectory tracking.
• To investigate the effectiveness of combining neural networks (NNs) with proportional integral derivative (PID) and fractional-order PID (FOPID) controllers.
• To develop a robust controller that minimizes integral-time-square error (ITSE) and control signal chattering.

Main Methods:

• Six control structures were designed: set-point-weighted PID (W-PID), W-FOPID, recurrent neural network (RNN)-like PID (RNNPID), RNN-like FOPID (RNN-FOPID), NN+PID, and NN+FOPID.
• The Zebra Optimization Algorithm (ZOA) was employed for optimizing controller parameters, minimizing ITSE.
• A novel objective function was introduced to reduce control signal chattering during tuning.
• A comparative robustness analysis was performed by varying initial conditions, disturbances, and model uncertainties.

Main Results:

• The NN+FOPID controller exhibited the best trajectory tracking performance among the evaluated controllers.
• The NN+FOPID controller achieved the minimum integral-time-square error (ITSE).
• This controller demonstrated superior robustness against variations in initial states, external disturbances, and parameter uncertainties.

Conclusions:

• The proposed NN+FOPID controller offers a significant improvement in trajectory tracking for robotic manipulators.
• Combining neural networks with fractional-order PID controllers provides enhanced robustness and performance in complex robotic systems.
• The developed control strategy effectively addresses challenges posed by nonlinear dynamics, disturbances, and uncertainties in robotic manipulator control.