基于遥感智能的中国分布式建筑光伏发电潜力评估方法及资源特性分析  

Intelligent Remote Sensing-based Evaluation and Analysis on China’s Distributed Building Photovoltaic Potentials

作　　者：陈晨 宋福龙 彭玲 陈德跃 陈伯煜 CHEN Chen;SONG Fulong;PENG Ling;CHEN Deyue;CHEN Boyu(Global Energy Interconnection Development and Cooperation Organization,Xicheng District,Beijing 100031,China;Aerospace Information Research Institute,Chinese Academy of Sciences,Chaoyang District,Beijing 100101,China;Key Laboratory of Smart Grid of Education Ministry,Tianjin University,Nankai District,Tianjin 300072,China)

机构地区：[1]全球能源互联网发展合作组织,北京市西城区100031 [2]中国科学院空天信息创新研究院,北京市朝阳区100101 [3]天津大学智能电网教育部重点实验室,天津市南开区300072

出　　处：《全球能源互联网》2025年第2期176-191,共16页Journal of Global Energy Interconnection

基　　金：国家自然科学基金专项项目(42341206);全球能源互联网集团有限公司科技项目(建筑光伏发电潜力评估方法及实证研究)。

摘　　要：近年来中国分布式建筑光伏快速发展,项目开发和并网规划都迫切需要更高精度和更大范围的建筑光伏资源评估方法。提出一种基于遥感智能的建筑光伏发电潜力评估方法,应用亚米级遥感卫星影像,分别设计基于自监督学习的屋顶识别算法和阴影识别测高算法,逐个识别测算得到全国建筑屋顶和建筑立面的光伏可开发面积,评估了装机潜力和年发电量,算法对建筑的平均识别精度达到87%。进一步分析了屋顶光伏最佳倾角分布及其对发电量的影响、建筑光伏资源分布影响因素和各省份分布式光伏开发前景分类。方法实现了“自下而上”批量化、自动化评估全国建筑光伏发电潜力,可灵活支撑建筑光伏项目规划设计、开发建设、并网规划等多层次需求,为全国建筑光伏有序开发提供了有效技术手段。In recent years,rapid development of distributed building-photovoltaics(PV)in China has heightened the need for more precise and expansive methods to evaluate building-PV potentials for project development and grid integration planning.This study introduces an intelligent remote sensingbased evaluation method for distributed building-PV potential.Using sub-meter resolution satellite images,algorithms to recognize rooftop and façade for individual buildings were developed based on self-supervised learning and building shadow recognition respectively.With the recognition accuracy reached 87%,the exploitable area,potential installed capacity and annual generation of nationwide building-PV were assessed.Furthermore,this paper analyzed the“optimized angles”of rooftop PV panels and their impact on PV power generations.Impact factors on building-PV resources were discussed,as well as province clusters classified by development prospects.This method achieved a“bottom-up”batch and automated assessment of building-PV potential,providing an effective technique to support building-PV project and power grid integration planning.

关 键 词：分布式建筑光伏 光伏发电潜力 卫星遥感 机器学习 自监督学习 

分 类 号：TM615[电气工程—电力系统及自动化] P236[天文地球—摄影测量与遥感]