基于LSTM网络模型的光伏发电功率短期预测系统
Short-term Power Prediction System for Photovoltaic Power Generation Based on LSTM Model
常振成 1游国栋 1肖梓跃 1李兴韫1
作者信息
- 1. 天津科技大学 电子信息与 自动化学院,天津 300222
- 折叠
摘要
光伏发电受天气因素影响,具有明显的间歇性和波动性特征.该文提出了一种基于LSTM网络模型的光伏发电功率短期预测方法,该方法以STM32单片机为控制核心,实时采集光照辐度、温度、相对湿度、风速等数据.利用相关系数法筛选相关度较高的因素,作为LSTM网络模型的输入变量,对未来光伏发电功率进行短期预测.MATLAB仿真实验结果表明,该文所提方法与其他预测模型相比具有较高的预测精度,在晴天与多云天气下预测的MAPE值分别为4.943％和4.997％,有利于我国电力系统的稳定运行和电网工作人员的调度.
Abstract
Photovoltaic power generation is affected by weather factors and has obvious intermittent and fluctuating characteristics.In this paper,a short-term prediction method of photovoltaic power generation based on LSTM network model is proposed,which uses STM32 microcontroller as the control core to collect data such as radiance,tempera-ture,relative humidity,and wind speed in real time.The correlation coefficient method is used to screen the factors with high correlation and use them as input variables of the LSTM network model to make short-term predictions of future photovoltaic power generation.The results of MATLAB simulation experiments show that the proposed method has high prediction accuracy compared with other prediction models,and the MAPE values predicted in sunny and cloudy weather are 4.943％and 4.997％respectively,which is conducive to the stable operation of China's power system and the dispatch of power grid staff.
关键词
STM32单片机/短时预测/LSTM网络模型/实时采集/光伏发电功率Key words
STM32 MCU/short-term forecasting/LSTM network model/real-time collection/photovoltaic power generation引用本文复制引用
基金项目
天津市应用基础与前沿技术研究计划(13JCZDJC29100)
天津市重点研发计划(17YFZCNC00230)
大学生创新创业训练计划(202310057101)
出版年
2024
NETL
NSTL
万方数据