Identification and control of resonant switch mode converters using neural networks

In this paper, the use of neural networks for identification and control of quasi-resonant converters (QRC) is investigated. The model of the DC-DC switching converter is implemented by means of a neural network emulator to identify the converter dynamics. The emulator reproduces the converter dynamic behavior accurately. Also a three-layer neural network controller is developed and applied to regulate the converter output voltage irrespective of changes in the input voltage or in the load. Adjusting the switching frequency of the QRC active switch regulates the output voltage. The QRC circuit controlled by the developed neural network is simulated, and the change in the output voltage is plotted versus the change in the load and the input voltage. The simulation results proved the feasibility of neural networks in controlling DC-DC converters. The accuracy of the proposed controller is also proved