Iterative Learning Control Without Resetting Conditions of an Algorithm Based on a Finite-Time Zeroing Neural...

Area of Science:

• Robotics
• Control Systems Engineering
• Artificial Intelligence

Background:

• Robotic manipulators often face challenges with trajectory tracking accuracy due to external disturbances.
• Existing iterative learning control (ILC) methods may require specific reset conditions, limiting their practical application in repetitive tasks.

Purpose of the Study:

• To design an advanced iterative learning control strategy for robotic manipulators that operates without reset conditions.
• To enhance trajectory tracking performance by mitigating external disturbances and improving convergence speed.

Main Methods:

• Development of a finite-time zeroing neural network (FTZNN) to effectively eliminate external disturbances.
• Proposal of a non-resetting iterative learning control without resetting conditions (NRCILC) framework that integrates FTZNN.
• The NRCILC-FTZNN automatically provides initial state values, removing the need for reset conditions in each iteration.

Main Results:

• The proposed NRCILC-FTZNN method achieved significant reductions in trajectory-tracking errors, with mean absolute error (MAE) improvements of 46.89% and 63.29% compared to other control schemes.
• Tracking errors converged to zero in fewer iterations than conventional methods.
• Simulations confirmed the robustness and effectiveness of the NRCILC-FTZNN scheme in robotic manipulator systems operating under disturbances.

Conclusions:

• The NRCILC-FTZNN is a highly effective control strategy for robotic manipulators, offering superior trajectory tracking performance.
• The elimination of reset conditions and the integration of FTZNN provide a robust and efficient solution for repetitive tasks with external disturbances.