Remarks on a Quaternion–Valued Neural Network–Based Direct Controller for Discrete–Time Nonlinear Systems

In this study, we explored the use of quaternion-valued neural networks (QVNNs) in controlling nonlinear systems. We utilised QVNN to design a direct controller, where its output serves as the plant input for the objective plant. Specifically, the QVNN was trained on the plant's input-output datasets to derive the inverse model of the plant. Computational experiments were conducted using a QVNN-based direct controller for a tracking task involving a discrete-time nonlinear plant. Experimental results demonstrate the viability of the QVNN for this control task with satisfactory accuracy, highlighting its potential for controlling nonlinear systems.