基于神经网络的PWM整流器矢量控制研究

建立电压型PWM整流器的一般数学模型，运用坐标变换技术得到其在两相旋转dq坐标系下的数学模型，并采用SVPWM整流器的双闭环控制系统。针对PI控制器参数整定困难的问题，利用神经网络的自学习功能，设计一种基于BP神经网络的PI控制器，实现PI控制器的参数自整定。最后利用MATLAB提供的电力系统工具箱构建PWM整流器的滞环控制系统和矢量控制系统的仿真模型进行仿真实验对比。仿真结果表明，基于BP神经网络的矢量控制系统超调量被抑制，系统鲁棒性增强。

Research on PWM rectifier vector control based on ANN

The general mathematical model of the voltage-type PWM rectifier was established, and the mathematical model of the rectifier under the two-phase rotating dq coordinate system was also estab- lished by using coordinate transformation technology. Dual closed loop control system of SVPWM rectifier was used. In light of the difficulties of tuning PI controller parameters, a PI controller based on BP neural network was designed by using the self-learning function of neural network for self-tun- ing of PI controller parameters. Finally, the power systems provided by MATLAB toolbox was used to construct the simulation model of PWM rectifier control hysteresis system and vector control sys- tem. The simulation results show that the overshoot of BP neural network-based vector control sys- tem is suppressed and the system robustness enhanced.