Exploiting more robust and efficacious deep learning techniques for modeling wind power with speed

Sound analyses of the nonlinear relationship between wind speed and power generation are crucial for the advancement of wind energy optimization. As an emerging artificial intelligence technology, deep learning has received growing attention from energy researchers for its outstanding ability to provide complex mappings. However, deep neural networks involve complex configurations, making it challenging to utilize them in practice. This paper assesses and presents a number of model-control techniques, categorized as model-oriented and data-oriented, to achieve more robust and efficacious deep neural networks for applications in the nonlinear modeling of wind power with wind speed. These carefully refined models are also compared with polynomials, simple neural networks, and not optimized deep networks with annual data of an Arctic wind farm. The results show that deep networks with sufficient parameter tunings, training optimizations, and modeling exhibit superior performance and generalization, thus possessing considerable advantages in wind energy engineering.