针对光伏发电量研究中传统单一模型预测误差大、特征数据少、深层神经网络模型出现梯度爆炸或消失的问题,该文提出一种基于多特征融合和XGBoost-LightGBM-ConvLSTM（XG-LG-CL）的短期光伏发电量预测模型。首先,分析影响光伏发电的相关因素,采用光伏领域特征融合和高阶特征融合方法将原有11个有效特征增加至62个有效特征;其次,分别建立XGBoost、LightGBM和ConvLSTM模型提取时空特征;最后,利用自适应权重法混合3种模型进行发电量预测。结果表明,该模型在光伏发电实测数据实验中,预测准确率为88.4%,与现有预测方法相比提升了3.1~8.6个百分点,可精确地预测光伏发电量,为电网稳定运行提供有效数据支撑。
Abstract
In this paper, a short-term PV power generation prediction model based on multi-feature fusion and XGBoost-LightGBM-ConvLSTM （XG-LG-CL） is proposed to solve the problems of large prediction error, few feature data and gradient explosion or disappearing of deep neural network model in the traditional single model. Firstly, the relevant factors affecting photovoltaic power generation are analyzed, and the original 11 effective features are increased to 62 effective features by photovoltaic field feature fusion and high-order feature fusion. Secondly, XGBoost, LightGBM and ConvLSTM models are established to extract the temporal and spatial features respectively. Finally, the adaptive weight method is used to mix the three models to predict the power generation. The results show that the prediction accuracy of the model is 88.4% in the experiment of measured data of photovoltaic power generation, which is improved by 3.1-8.6 percentage points compared with the existing prediction method. The model can accurately predict photovoltaic power generation and provide effective data support for the stable operation of power grid.
关键词
光伏发电 /
数据挖掘 /
特征融合 /
XGBoost /
LightGBM /
ConvLSTM
Key words
photovoltaic power generation /
data mining /
feature fusion /
XGBoost /
LightGBM /
ConvLSTM
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 唐雅洁, 林达, 倪筹帷, 等. 基于XGBoost的双层协同实时校正超短期光伏预测[J]. 电力系统自动化, 2021, 45(7): 18-27.
TANG Y J, LIN D, NI C W, et al.XGBoost based bi-layer collaborative real-time calibration for ultra-short-term photovoltaic prediction[J]. Automation of electric power systems, 2021, 45(7): 18-27.
[2] ZHOU H X, ZHANG Y J, YANG L F, et al.Short-term photovoltaic power forecasting based on long short term memory neural network and attention mechanism[J]. IEEE access, 2019, 7: 78063-78074.
[3] 张静, 褚晓红, 黄学安, 等. 一种基于加权马尔科夫链修正的SVM光伏出力预测模型[J]. 电力系统保护与控制, 2019, 47(19): 63-68.
ZHANG J, CHU X H, HUANG X A, et al.A model for photovoltaic output prediction based on SVM modified by weighted Markov chain[J]. Power system protection and control, 2019, 47(19): 63-68.
[4] 朱琳琳, 钟志峰, 严海, 等. 一种新的光伏发电预测模型设计[J]. 太阳能学报, 2016, 37(1): 63-68.
ZHU L L, ZHONG Z F, YAN H, et al.A new design of photovoltaic power generation forecasting model[J]. Acta energiae solaris sinica, 2016, 37(1): 63-68.
[5] 王洪坤, 葛磊蛟, 李宏伟, 等. 分布式光伏发电的特性分析与预测方法综述[J]. 电力建设, 2017, 38(7): 1-9.
WANG H K, GE L J, LI H W, et al.A review on characteristic analysis and prediction method of distributed PV[J]. Electric power construction, 2017, 38(7): 1-9.
[6] 崔佳豪, 毕利. 基于混合神经网络的光伏电量预测模型的研究[J]. 电力系统保护与控制, 2021, 49(13): 142-149.
CUI J H, BI L.Research on photovoltaic power forecasting model based on hybrid neural network[J]. Power system protection and control, 2021, 49(13): 142-149.
[7] 段美玲, 潘巨龙. 基于Lasso-LGB的老人跌倒检测算法研究[J]. 中国计量大学学报, 2021, 32(1): 67-73, 117.
DUAN M L, PAN J L.Research on elderly fall detection algorithm for the elderly based on Lasso-LGB[J]. Journal of China University of Metrology, 2021, 32(1): 67-73, 117.
[8] 贾睿, 杨国华, 郑豪丰, 等. 基于自适应权重的CNN-LSTM&GRU组合风电功率预测方法[J]. 中国电力, 2022, 55(5): 47-56, 110.
JIA R, YANG G H, ZHENG H F, et al.Combined wind power prediction method based on CNN-LSTM&GRU with adaptive weights[J]. Electric power, 2022, 55(5): 47-56, 110.
[9] 光伏电站人工智能运维大数据处理分析[DB/OL].[2018-07-13].https://www.datafountain.cn/competitions/303/datasets.html.
Big data processing and analysis of artificial intelligence operation and maintenance of power station.[DB/OL].[2018-07-13].https://www.datafountain.cn/competitions/303/datasets.html.
[10] 彭曙蓉, 郑国栋, 黄士峻, 等. 基于XGBoost算法融合多特征短期光伏发电量预测[J]. 电测与仪表, 2020, 57(24): 76-83.
PENG S R, ZHENG G D, HUANG S J, et al.Multiple-feature short-term photovoltaic generation forecasting based on XGBoost algorithm[J]. Electrical measurement and instrumentation, 2020, 57(24): 76-83.
[11] 王坤. 考虑多重不确定性因素的光伏出力预测研究[D].北京: 华北电力大学,2013.
WANG K.Output prediction research of photovoltaic system considering multiple uncertainty factors[D]. Beijing: North China Electric Power University, 2013.
[12] 吕学梅, 孙宗义, 曹张驰. 电池板温度和辐射量对光伏发电量影响的趋势面分析[J]. 可再生能源, 2014, 32(7): 922-927.
LYU X M, SUN Z Y, CAO Z C.A trend surface analysis on effect of battery plate temperature and radiation levels on PV power output[J]. Renewable energy, 2014, 32(7): 922-927.
基金
宁夏自然科学基金（2023AAC02011）; 宁夏重点研发项目（2021BEE03020）
PDF(2349 KB)
