上海交通大学学报

上海交通大学学报

• •    

基于神经网络的LCL型单相并网逆变器稳定性分析（网络首发）

  

  1. 1. 中国电力科学研究院有限公司南京分院；2. 武汉大学电气与自动化学院；3. 综合能源电力装备及系统安全湖北省重点实验室；4. 国网陕西省电力有限公司电力科学研究院
  • 基金资助:
    国家电网有限公司科技项目(4000-202355078A-1-1-ZN)；

Stability Analysis of LCL Grid-Connected Inverter Based on Neural Network

  1. （1. Nanjing Branch, China Electric Power Research Institute Co., Ltd., Nanjing 210009, China；2. School of Electrical Engineering and Automation, Wuhan University, Wuhan 430072, China； 3. Hubei Key Laboratory of Power Equipment and System Security for Integrated Energy, Wuhan 430072, China；4. Electric Power Research Institute, State Grid Shaanxi Electric Power Co., Ltd., Xi’an 710100, China）

PDF (PC)

15

摘要： LCL型并网逆变器的参数不确定性严重影响新能源发电的电能质量，因此需要对参数摄动情况下逆变器的稳定情况进行分析。面对上述问题，该文首先建立LCL型单相并网逆变器状态空间模型，提出基于参数模型的神经网络建模方法。通过参数特征分析，得到基于参数分布的训练数据集，在神经网络中对数据集进行训练，输出稳定性判别结果。最后，通过MATLAB/Simulink仿真与实验平台验证所提基于神经网络分析逆变器稳定性方法的有效性。

关键词: LCL型单相并网逆变器, 稳定性分析, 神经网络, 反向传播, 二分类

Abstract: The parameter uncertainty of LCL-type grid-connected inverter can significantly affect the power quality of renewable energy, so it is necessary to analyze the stability of the inverter under the parameter inception. Facing the above problems, the paper firstly establishes the state space model of LCL-type single-phase grid-connected inverter and proposes the neural network modeling method based on the parameter model. Through the parameter characterization, the training dataset based on the parameter distribution is obtained. This dataset is trained in the neural network to output the stability discrimination results. Finally, the effectiveness of the proposed method is verified by MATLAB/Simulink simulation and the experimental platform.

Key words: LCL type single-phased grid connected inverter, stability analysis, neural network, back propagation, binary classification

中图分类号: