Research on the Model of the PV Grid-Connected Power Generation System Based on the Neural Network

• Published in: 2023 IEEE International Conference on Power Science and Technology (ICPST) pp. 528 - 533
• Main Authors: Liu, Kezhen, Mao, Yumin, Zhang, Lanbin, Lin, Zheng, Yang, Chunhao, Pu, Wei
• Format: Conference Proceeding
• Language: English
• Published: IEEE 05.05.2023
• Subjects: Adaptation models; BP neural network; dynamic characteristics; electromechanical transient; Neural networks; Power system dynamics; Power system stability; Predictive models; PV grid connection; Renewable energy sources; Voltage fluctuations
• DOI: 10.1109/ICPST56889.2023.10165366

Under the "double carbon" goal, the power system gradually presents the "double high" characteristics of the high proportion of renewable energy and power electronic equipment. Renewable energy represented by photovoltaic (PV) is connected to the power grid on a large scale, which significantly changes the load characteristics and transient stability of the power system. To study the dynamic characteristics of the PV grid-connected power generation system, it is urgent to establish a dynamic model that meets the actual operating conditions, but most of the models are complex and have many parameters to be identified. At the same time, the model parameters are not unique with the change of the PV penetration rate and voltage dip depth, lacking certain engineering practicability. Therefore, based on the electromechanical transient characteristics of the PV grid-connected power generation system, the back propagation (BP) neural network is used to model the real and imaginary parts of the current at the grid-connected point, respectively, and the model of the PV grid-connected power generation system based on the BP neural network is proposed, which does not need to predict the mathematical equations of the mapping relationship between the input and output, and has the simple structure and strong nonlinear implicit capability. Finally, the IEEE14 node system model is built in MATLAB/Simulink to verify the accuracy and applicability of the model.