广东电力2025，Vol.38Issue(6)：19-29,11.DOI:10.3969/j.issn.1007-290X.2025.06.003
基于二维建模和BP神经网络回归法的区域屋顶光伏资源容量测算方法
Measurement and Calculation Method for Regional Rooftop Photovoltaic Resource Capacity Based on Two-dimensional Modeling and BP Neural Network Regression
摘要
Abstract
High-proportion rooftop photovoltaics,characterized by large total capacity,scattered distribution,and strong randomness,have increasingly become a significant power source that can't be ignored.Their integration into the distribution networks brings uncertainties,making the accurate and efficient estimation of the distributed rooftop photovoltaic capacity a crucial issue in the power industry.In order to mitigate the uncertainties caused by the installation of photovoltaic panels and weather conditions,this paper proposes a method combining two-dimensional modeling and back propagation(BP)neural network prediction for estimating the capacity of regional rooftop photovoltaic resources.Firstly,to address the issue of accuracy affected by roof shapes in rooftop areas,the convolutional block attention module(CBAM)attention module is introduced into the traditional DeepLab V3+neural network.By means of the noise reduction functions of the spatial attention modules and channel attention modules,it is able to enhance the ability of neural network to filter effective information such as roof types and shapes.Secondly,by calculating the north-south gap of photovoltaic panels and conducting iterative translation operations,the coverage of photovoltaic module layouts is improved without shadow obstructions between photovoltaic panels and dispensing with optimization solvers.To reduce uncertainties caused by weather factors,the BP neural network is used to predict the solar radiation intensity of the research object,obtain peak hours,and calculate annual electricity generation capacity.Finally,through simulation examples,the maximum installed capacity and annual electricity generation capacity of the study areas are calculated,proving the effectiveness of the proposed algorithm.关键词
屋顶光伏/光伏组件/空间布置/容量测算/太阳辐射预测Key words
rooftop photovoltaic/photovoltaic module/spatial layout/capacity calculation/solar radiation forecasting分类
信息技术与安全科学引用本文复制引用
欧阳森,康澜,彭志豪..基于二维建模和BP神经网络回归法的区域屋顶光伏资源容量测算方法[J].广东电力,2025,38(6):19-29,11.基金项目
国家自然科学基金项目(52177085) （52177085）
