A PI Self-Tuning Method for Three-Phase PWM Rectifiers Based on Stability-and-Dynamics- Constrained Fuzzy Backpropagation Neural Network

In three-phase PWM rectifiers, variations in the dc voltage reference or load can affect the stability and dynamics of the system. Traditional proportional-integral (PI) control has weak adaptability, as its parameters cannot be adjusted in real-time according to the operating state, potentially leading to poor system stability and slow dynamic response. To address these challenges and the complexity of existing PI self-tuning methods, this article proposes a PI self-tuning method based on stability-and-dynamics-constrained fuzzy backpropagation neural network (SDC-FBPNN) for three-phase pulsewidth modulation (PWM) rectifiers. The effect of the right-half-plane zero is considered, and a series compensation method is adopted to convert the non-minimum phase system into a minimum phase system. On this basis, the phase margin and crossover frequency are used as constraints for stability and dynamic performance, respectively. Consequently, PI parameters can be adjusted by the SDC-FBPNN in real-time according to the operation conditions. Under abrupt variations in voltage reference and load, the proposed method significantly reduces voltage fluctuations and shortens the settling time while ensuring system stability. Finally, a 2 kW prototype was built to validate the feasibility and effectiveness of the proposed PI self-tuning method.