Non-mechanistic Model of Doubly Fed Induction Generators on Artificial Neural Networks

Doubly fed induction generators (DFIG) are widely used in constant-frequency wind power generation systems. The traditional mathematical mechanism model of DFIGs presents high-dimensional complexity in complex application scenarios such as wind power grid integration systems. This paper proposes a non-mechanical modeling method for DIFG based on Recurrent Neural Network (RNN) neural networks. This method collects data from DFIG grid integration systems, constructs input and output features, and uses the RNN algorithm to establish the nonlinear mapping relationships between features, resulting in a non-mechanical model of the DFIG grid integration system. This model can achieve high-precision prediction of various electrical quantities in DFIG grid integration systems. Experimental verification demonstrates the feasibility of this method, indicating that it can model different features under various working conditions and achieve high accuracy.