abstract
- In this article, the idea is to implement an integral to the operator of the adaptation algorithm of a Feed-forward Neural Network (FNN) trained online to reduce chattering in trajectory tracking at the end effector of a robot with a flexible link. The controller structure is divided into a linear control part that controls the angular position and a data-based controller to controls the nonlinear part of the system. The neural network (NN) is trained online to approximate the dynamics of the flexible link. The controller's performance in trajectory tracking and chattering reduction is measured using the Mean Square Error (MSE). The controller is compared against an NN-based controller without the integral. The results show that trajectory tracking error and chattering are reduced by using an integrated NN training law compared to non-integrated NN. © 2023 IEEE.