Field programmable gate array-based neural network control strategy for computer power supply applications

A Multiple Output Switched Mode Power Supply (MOSMPS) with Artificial Neural Network (ANN) control for Power Quality (PQ) improvement and output voltage regulation is proposed for personal computers in this paper. The proposed MOSMPS eliminates the conventional diode bridge rectifier at the input by incorporating a bridgeless converter to minimize conduction losses and improve thermal management. The bridgeless converter is modeled to operate in Discontinuous Conduction Mode (DCM) for advantages like improved Power Factor (PF), zero-current switching of power switches, and several sensors to implement the control strategy effectively. The proposed ANN controller is simulated and experimentally verified for various source and load conditions. An FPGA processor is implemented to experimentalize ANN control. The efficacy of the proposed configuration with an ANN controller is validated in terms of PQ indices by comparing the simulation and experimental results with those of another topology and conventional controller.