Integration of Wind Resource Assessment through Multi-scale Coupling Method

Abstract

Assessing wind energy resources holds paramount significance in addressing the conventional energy crisis and ensuring the sustained, rapid development of the national economy. The study of wind resource assessment serves as a critical indicator used to evaluate the construction and operation of large-scale wind power projects, which holds crucial guiding significance in selecting wind farm models, equipment procurement, and operational revenue optimization. Therefore, aimed at enhancing the precision of wind resource assessment in complex terrains, this paper presents a method utilizing machine learning coupled with multi-scale numerical simulation. Firstly, wind profile simulation results are extracted from the mesoscale Weather Research and Forecasting (WRF) model. Subsequently, to integrate multi-scale temporal and spatial data, these mesoscale wind profiles are fitted using multi-task learning(MTL). Finally, the turbulence is derived by a periodical running precursor on Computational Fluid Dynamics (CFD), utilizing the fitted outcomes as input boundary conditions for microscale simulations, complex terrain is incorporated to enhance the accuracy of wind resource assessment. The results show that the proposed method can improve the assessment accuracy of wind resources of wind farms by mesoscale data, which is valuable to a certain extent in real-world applications.