Physics-Informed Neural Network for Transient Modeling of Large-Scale Wind Farm

The complexity of transient characteristics in large-scale wind farms (WF) hinders the application of machine learning algorithms. This paper proposes a neural network-based learning method to model the transient characteristics of large-scale WF. The complexity of the physical information library is simplified through an iterative algorithm. The dynamic library obtained based on neural networks can induce the machine learning framework to rapidly learn the transient characteristics of large-scale WF. Moreover, there is no need for excessive mechanistic analysis and speculation regarding the transient behavior of WF. The effectiveness of the proposed method is verified in the simulation model of a WF.