Three-Phase PWM Rectifier Control: Enhanced Direct Power Control with Neural Networks from Theory to Superior Reality Performance

This paper proposes an innovative enhancement to conventional Direct Power Control (DPC) using Neural Network Control (NNC), presenting an effective alternative control method for three-phase Pulse Width Modulation (PWM) rectifiers.Traditional DPC techniques struggle with dynamic non-linearity and system uncertainties, leading to issues such as limited resolution, lack of adaptability, interpolation errors, sensitivity to noise, overshooting, and distorted grid currents.To address these challenges, the proposed NNC algorithm replaces the PI controller of the DC link voltage, hysteresis comparators for active and reactive power, and the lookup table.The NNC algorithm is distinguished by its nonlinear mapping capabilities and real-time parallel processing.The effectiveness of the proposed approach was validated through experimental and simulation results, using the DSpace DS1103 card along with MATLAB and Control Desk software.The Total Harmonic Distortion (THD) of the grid current in simulation and experimental results is recorded as 1.20% and 4.68%, respectively demonstrating significant improvements in overall system performance in both steady and transient states, proving the efficiency, robustness, and effectiveness of the NNC algorithm compared to traditional methods.