FAULT-TOLERANT GRID SYNCHRONIZATION OF THREE-PHASE PWM RECTIFIERS USING NEURAL NETWORK

This work focuses on the essential role of synchronous angle estimation in the conversion of three-phase AC to DC voltage. Accurately determining the synchronous angle is crucial for achieving optimal phasing in the abc/d-q transformation, aligning the synchronous reference d-q with the phase “a” voltage vector. This alignment ensures the efficient operation of the inner current loops in the d-q synchronous reference. However, estimating the synchronous angle becomes challenging during AC grid faults. Previous studies have employed phase-locked loops (PLLs) with specialized filters to handle parasitic harmonics. In contrast, this research proposes using neural networks to recognize fault types and estimate the phase with minimal perturbations. Comparative analyses of conventional PLL-based and neural network-based estimates of the converter bridge's output performance under fault conditions are presented using MATLAB/Simulink/PLECS.