Open AccessNeural Network Model of Wind Farm Based on DFIGs
Author(s) -
Bingtao Guo,
Bin Xie,
Li Zhan,
Xu Cai
Publication year - 2019
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1346/1/012030
Subject(s) - induction generator , control theory (sociology) , artificial neural network , wind speed , power (physics) , wind power , voltage , ac power , nonlinear system , rotor (electric) , computer science , function (biology) , engineering , artificial intelligence , physics , electrical engineering , control (management) , quantum mechanics , meteorology , evolutionary biology , biology
Due to the complexity of wind farms for power flow calculation in power system, it is difficult to accurately describe equivalent models in accordance with traditional methods such as weighted average parameters. However, the merit of neural network (NN) is able to mimic nonlinear and complicated function with multiple inputs and multiple outputs (MIMO). The back-propagation (BP) NN is adopted to substitute the wind farm based on doubly-fed induction generators (DFIG) in this paper. To do this, a simple BP model is established to be equivalent to single DFIG. The training data of the NN is obtained by simulation of actual DFIG, in which the wind speed, voltages of DFIGs, optimal power captured strategy are used as inputs and the active and reactive power are as outputs. Furthermore, a multilayer BP model is built and trained to mimic the wind farm based on DFIGs using simulation results of the actual 34 DFIGs, 102 MW electrical network with statistical wind speed and rotor-side voltage. The simulation results show that the BP NN can be used to model the wind farm, simplify the power flow calculation in the power system and be of high accuracy.