Real-time implementation of an on-line trained neural network controller for power electronics converters

Since power electronics converters behave nonlinearly, conventional control strategies such as PID are incapable of obtaining good dynamical performance. This paper addresses implemention of on-line trained neural networks for power electronics converters. A PWM boost converter is used as an example. Real-time implementation of the neural networks is accomplished by using a powerful digital signal processor. The converter is operated as a power amplifier and a power regulator. Both computer simulation and experimental results show that good dynamical performance can be obtained