电力系统及其自动化学报2025，Vol.37Issue(10)：23-31,9.DOI:10.19635/j.cnki.csu-epsa.001569
基于神经网络的中点钳位型三电平并网逆变器模型预测控制
Model Predictive Control of Neutral-point-clamped Three-level Grid-connected Inverter Based on Neural Network
摘要
Abstract
Aimed at the problems of poor robustness of model predictive control(MPC)parameters and large output cur-rent ripple in a neutral-point-clamped three-level grid-connected inverter,an MPC strategy based on radial basis func-tion(RBF)neural network is proposed.First,the weighted average current control principle is used to reduce the sys-tem from third-order to first-order,and its simplified prediction model is established.Second,the disturbance caused by device aging,parameter perturbation and parameter deviation coupling is regarded as lumped disturbance,which is estimated by using the universal approximation characteristics of RBF neural network,and the prediction model of weighted average current is compensated in real time.Finally,the stability of the proposed control strategy is proved by the rigorous Lyapunov theory.Experimental results show that the proposed strategy significantly enhances the system ro-bustness and effectively suppresses the output current ripple.关键词
中点钳位型三电平并网逆变器/模型预测控制/参数鲁棒性/径向基神经网络/加权平均电流控制/Lyapunov 理论Key words
neutral-point-clamped three-level grid-connected inverter/model predictive control(MPC)/parameter ro-bustness/radial basis function(RBF)neural network/weighted average current control/Lyapunov theory分类
信息技术与安全科学引用本文复制引用
徐宏强,何国锋,赵俊苗,张洛花,李睿..基于神经网络的中点钳位型三电平并网逆变器模型预测控制[J].电力系统及其自动化学报,2025,37(10):23-31,9.基金项目
国家自然科学基金资助项目(52177187) （52177187）
河南省高等学校重点科研项目计划资助(23B413005). （23B413005）
