基于DRNN自整定QPR-DMRC控制的并网逆变器技术

针对常规并网逆变器中电网电压会出现畸变及传统控制器动态响应差等问题,通过分析准比例谐振和传统重复控制策略的优缺点,提出了一种基于对角递归神经网络的改进型QPR-双模重复控制(DMRC)复合控制器并给出其控制算法,DRNN采用LM算法,利用DRNN参数自整定技术,对改进型QPR-DMRC控制器参数进行在线整定,该方法既能够有效地对奇、偶次谐波进行抑制,同时解决了QPR控制器参数整定困难等问题。采用Matlab/Simulink进行仿真研究,结果表明该方法能有效地降低系统谐波总畸变率,提高了系统的抗干扰能力,实现逆变器无静差稳定运行。

Grid-connected Inverter Technology Based on DRNN under Self-turning QPR-DMRC Control

Considering problems such as grid voltage distortion in the conventional grid-connected inverter and poor dy?namic response of the traditional controller,an improved quasi-proportional resonance(QPR)-double mode repetitive control(DMRC)compound controller based on diagonal recurrent neural network(DRNN)is proposed by analyzing the advantages and disadvantages of QPR and the traditional repetitive control(RC)strategy,and its control algorithm is given.The DRNN adopts the Levenberg-Marquardt(LM)algorithm.Moreover,by using the parameter self-tuning tech?nology of DRNN to adjust the parameters of the improved QPR-DMRC controller online,this method not only effective?ly suppresses the odd-and even-order harmonics,but also solves problems including the difficulty in tuning the QPR controller’s parameters.Simulations are conducted using Matlab/Simulink,and results show that the proposed method can effectively reduce the system’s total harmonic distortion rate,improve its anti-jamming capability,and realize the inverter’s stable operation with zero steady-state error.