期刊
APPLIED ENERGY
卷 250, 期 -, 页码 283-291出版社
ELSEVIER SCI LTD
DOI: 10.1016/j.apenergy.2019.04.113
关键词
Deep learning; Neural networks; Solar energy; Urban energy; Solar potential mapping
资金
- National Natural Science Foundation [51678261]
- State Key Laboratory of Subtropical Building Science [20172808]
- Wuhan Urban and Rural Construction Committee [201726]
This study presents a novel approach to detect the city-wide solar potential which utilizes image segmentation with deep learning technology unlike traditional methods. In order to study the solar energy potential in the urban scale, there exists a requirement to quantify the roof area of buildings which are available to receive solar radiation, calculate the total solar radiation obtained within the region based on the meteorological conditions, and determine the total solar energy potential with carbon emissions savings and the economic recovery period. However, obtaining the overall roof area of a city is an existing difficulty when considering the quantification of solar potential in the urban scale. This study utilizes the U-Net of deep learning technology, and a large range of satellite maps to identify the building roof, in order to estimate the city's solar potential. This research established that the urban roofs of Wuhan have an annual photovoltaic electricity generation potential of 17292.30 x 10(6) kWh/year.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据