ANN-based Load Identification and Control of a Power AC Voltage Regulator

M. Akherraz, A. Gastli, and S.M. Al-Alawi


AC regulator, triggering angle, extinction angle, artificial neural network, modelling, MATLAB/SIMULINK


The authors present the application of artificial neural network techniques to the load identification and control of AC voltage regulators. The simplicity and cost-effectiveness of these SCR- based AC-AC converters are overshadowed by the complexity of their control logic. With unknown and/or variable inductive loads, the SCR’s triggering and extinction angles depend on the load parameters in a highly nonlinear fashion. The ANN techniques are proven suitable for parameter identification and control of such nonlinear systems. A network topology with four input nodes, two output nodes, and two hidden layers is proposed. The feedforward network is trained with the back-propagation algorithm using MATLAB Neural Network Toolbox. The AC regulator, the load, and the ANN models are implemented with SIMULINK, which allows real-time visualization, adjustment, and control of all the input parameters. The obtained simulation results confirm the adequacy of the proposed ANN approach to identify the load parameters with high accuracy, and thus provide the AC regulator’s controller with the appropriate triggering interval.

Important Links:

Go Back