基于云平台的光伏电站监测及功率预测系统  

Monitoring and power prediction system of photovoltaic power station based on cloud platform

作　　者：海涛[1] 肖健伦 韦文[2] 张川 陆猛 HAI Tao;XIAO Jian-lun;WEI Wen;ZHANG Chuan;LU Meng(College of Electrical Engineering,Guangxi University,Nanning 530004,China;China Tobacco Guangxi Industrial Co.,Ltd.,Nanning 530001,China;Holgus Zhengtai Technology Service Co.,Ltd.,Holgus 835221,China)

机构地区：[1]广西大学电气工程学院,南宁530004 [2]广西中烟工业有限责任公司,南宁530001 [3]霍尔果斯正泰科技服务有限公司,新疆霍尔果斯835221

出　　处：《桂林理工大学学报》2022年第1期249-255,共7页Journal of Guilin University of Technology

基　　金：国家自然科学基金项目(51867003);广西科技计划项目(桂科AB16380193);南宁市西乡塘区科学研究与技术开发计划项目(2019035502)。

摘　　要：针对传统的嵌入式预测系统容易受到现场条件与自身硬件的影响,光伏发电系统监测距离短且功耗高、数据采集精度低、嵌入式预测系统稳定性差等问题,设计了以物联网云平台为主体框架,以LoRa为主要通信技术的光伏电站监测系统。通过在云端部署服务,使用GA-Elman神经网络模型,达到降低现场硬件成本与提高整体预测精度的目的。以湖北武当湖光伏电站为实验对象,使用本系统对电站内装机容量为19.9 kW的光伏矩阵进行实时监测。实验证明:该系统能长期可靠运行,实时监测光伏电站的各项数据,功率预测精度高,扩展性强。Traditional embedded prediction system is susceptible to the influence of field conditions with its own hardware.The current photovoltaic power generation system has a short monitoring distance,high power consumption,low data acquisition accuracy,and poor stability of the embedded prediction system.A photovoltaic power station monitoring system is designed with the internet of things cloud platform as the main framework and LoRa as the main communication technology.By deploying services in the cloud and using GA-Elman neural network model,this system can reduce the cost of field hardware and improve the overall prediction accuracy.This system is tested to monitor the power-on capacity of 19.9 kW photovoltaic matrix in Wudang Lake photovoltaic power station in Hubei.Experiments show that the system can run reliably for a long time,and monitor every data of photovoltaic power station in real time,with high accuracy and scalability in power prediction.

关 键 词：物联网 云平台 LoRa 神经网络 功率预测 

分 类 号：TM63[电气工程—电力系统及自动化] TM76