Offshore wind farm wake modelling using deep feed forward neural networks for active yaw control and layout optimisation

Abstract

Abstract Offshore wind farm modelling has been an area of rapidly increasing interest over the last two decades, with numerous analytical as well as computational-based approaches developed, in an attempt to produce designs that improve wind farm efficiency in power production. This work presents a Machine Learning (ML) framework for the rapid modelling of wind farm flow fields, using a Deep Neural Network (DNN) neural network architecture, trained here on approximate turbine wake fields, calculated on the state-of-the-art wind farm modelling software FLORIS. The constructed neural model is capable of accurately reproducing single wake deficits at hub-level for a 5MW wind turbine under yaw and a wide range of inlet hub speed and turbulence intensity conditions, at least an order of magnitude faster than the analytical wake-based solution method, yielding results with 1.5% mean absolute error. A superposition algorithm is also developed to construct flow fields over the whole wind farm domain by superimposing individual wakes. A promising advantage of the present approach is that its performance and accuracy are expected to increase even further when trained on high-fidelity CFD or real-world data through transfer learning, while its computational cost remains low.