Resource allocation network based neural PID adaptive control for generator excitation system

In this paper, a neural PID adaptive controller based on the identification of resource allocation network (RAN) was proposed for the generator excitation system, which not only has the ability of neural network such as powerful nonlinear mapping and self-learning, but also can adjust the PID controller performance. The controller dynamically increases the number of hidden nodes through the learning samples, building the model online and dynamically adjusting the PID parameters to achieve the system output tracking input. Simulation results show that this method is better to stabilize the static and dynamic terminal voltages of the generator compared to the conventional PID control.