Estimation of the influences of spatiotemporal variations in air density on wind energy assessment in China based on deep neural network

This study proposes a deep neural network-based wind energy assessment method, aiming to system-atically evaluate the effects of the spatiotemporal variations in air density on wind energy assessment in China. On this basis, the spatiotemporal patterns of air density are modelled on a high-spatial-resolution scale (1000 m x 1000 m). The influence of the spatiotemporal distribution characteristics of air density on the wind energy production is quantified, and the spatiotemporal variability of the corresponding wind energy production is estimated. The results demonstrate that changes in air density during the year present typical periodic characteristics. The mean air density value in January is 1.079 kg/m(3), the highest throughout the year. The difference in mean air density between cold and warm seasons in the study area shows a decreasing law of higher in the northeast and lower in the southwest. When the elevation is less than 3500 m, it reaches 5.06%. The observed spatiotemporal variability in annual energy production exhibits a distinct seasonal cycle, with the highest production appears in spring (2.968 GWh/yr). The total annual energy production in the cold season is 16.08 GWh/yr, whereas the annual energy pro-duction decreases by higher than 23.46% when it comes to the warm season. (c) 2021 Elsevier Ltd. All rights reserved.

Automated summary

This study proposed a deep neural network-based method to assess wind energy in China, focusing on the effects of spatiotemporal variations in air density. The research found typical seasonal characteristics in air density changes and demonstrated the impact of air density distribution on wind energy production.