[1]刘兴霖,黄 超,王 龙,等.基于聚类和 LSTM 的光伏功率日前逐时鲁棒预测[J].计算机技术与发展,2023,33(03):120-126.[doi:10. 3969 / j. issn. 1673-629X. 2023. 03. 018]
　LIU Xing-lin,HUANG Chao,WANG Long,et al.Clustering and LSTM-based Robust Day-ahead Hourly Forecasting of Photovoltaic Power[J].,2023,33(03):120-126.[doi:10. 3969 / j. issn. 1673-629X. 2023. 03. 018]
点击复制

基于聚类和 LSTM 的光伏功率日前逐时鲁棒预测()
分享到：

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
33
期数:
2023年03期
页码:
120-126
栏目:
人工智能
出版日期:
2023-03-10

文章信息/Info

Title:
Clustering and LSTM-based Robust Day-ahead Hourly Forecasting of Photovoltaic Power
文章编号:
1673-629X(2023)03-0120-07
作者:
刘兴霖1; 2 ; 黄 超1; 2 ; 王 龙1; 2 ; 罗 熊1; 2
1. 北京科技大学 顺德研究生院,广东 佛山 528399;
2. 北京科技大学 计算机与通信工程学院,北京 100083
Author(s):
LIU Xing-lin1; 2 ; HUANG Chao1; 2 ; WANG Long1; 2 ; LUO Xiong1; 2
1. Shunde Graduate School,University of Science and Technology Beijing,Foshan 528399,China;
2. School of Computer and Communication Engineering,University of Science and Technology Beijing,Beijing 100083,China
关键词:
光伏发电预测; 长短期记忆神经网络; K-means 聚类; Huber 损失函数; 鲸鱼优化算法
Keywords:
photovoltaics power generation forecasting;  long - short - term memory neural network;  K - means clustering;  Huber lossfunction; whale optimization algorithm
分类号:
TP183
DOI:
10. 3969 / j. issn. 1673-629X. 2023. 03. 018
摘要:
太阳能作为具有高可用性且用之不竭的清洁能源,被认为是最有前途的能源替代品之一。 光伏是最广泛使用的太阳能技术。 然而,由于太阳能的间歇性,光伏发电具有不确定性。 随着全球光伏装机容量的不断提升,光伏功率预测的准确性对于电网管理和电力调度至关重要。 该文提出一种基于 K-means 聚类分析和长短期记忆神经网络( long -short-term memory,LSTM) 的光伏发电功率日前逐时鲁棒预测方法。 首先采用 K-means 算法以日前天气预报数据为特征将光伏数据分为晴空天气类型和阴雨天气类型,再针对相应类型数据建立基于长短期记忆神经网络算法的预测模型。 同时,为增强预测模型的鲁棒性,选择具有强鲁棒性的 Huber 损失函数用于模型训练,并选择计算简单且收敛速度快的鲸鱼优化算法对 Huber 损失函数中的超参数进行优化。 将所提出的预测方法与其他方法进行预测性能的比较,结果表明,提出的方法获得了较高的预测精度。
Abstract:
Due to its inexhaustible nature and friendliness to environment, solar energy is considered to be one of the most promisingenergy alternatives to fossil fuels. Photovoltaics ( PV) is the most widely used technology to make use of solar energy. However,PVpower generation is uncertain due to the intermittent nature of solar energy. With the increasing installation of PV power plants,accurateforecasting of PV power generation is critical to grid management and power dispatch. We propose a robust day-ahead hourly PV powergeneration forecasting method based on K-means clustering algorithm and long-short-term memory ( LSTM) neural network. The K-means algorithm is firstly used to classify PV data into clear sky weather type and rainy weather type based on day-ahead forecasting ofweather variables,and then LSTM-based deep learning forecasting models are developed for the corresponding types of data. In order toenhance the robustness of the forecasting model, the Huber loss function is selected for model training, and the whale optimizationalgorithm ( WOA) is selected to optimize the hyperparameter in the Huber loss function. The forecasting performance of the proposedmethod is compared with benchmarks. The results show that the proposed method can achieve higher forecasting accuracy.

相似文献/References:

[1]蒋锐鹏,姑丽加玛丽·麦麦提艾力,安丽娜.基于长短期记忆神经网络的手写数字识别[J].计算机技术与发展,2020,30(02):94.[doi:10. 3969 / j. issn. 1673-629X. 2020. 02. 019]
　JIANG Rui-peng,Gulijiamali·MAIMAITIAILI,AN Li-na.Handwritten Number Recognition Based on Long Short-term Memory Neural Network[J].,2020,30(03):94.[doi:10. 3969 / j. issn. 1673-629X. 2020. 02. 019]
[2]甄俊涛,刘 臣.高维数据多标签分类的食品安全预警研究[J].计算机技术与发展,2020,30(09):109.[doi:10. 3969 / j. issn. 1673-629X. 2020. 09. 020]
　ZHEN Jun-tao,LIU Chen.Research on Food Safety Early Warning of Multi-label Classification of High Dimensional Data[J].,2020,30(03):109.[doi:10. 3969 / j. issn. 1673-629X. 2020. 09. 020]
[3]刘高军,王小宾.基于 CNN+LSTMAttention 的营销新闻文本分类[J].计算机技术与发展,2020,30(11):59.[doi:10. 3969 / j. issn. 1673-629X. 2020. 11. 011]
　LIU Gao-jun,WANG Xiao-bin.Marketing News Text Classification Incorporating CNN+LSTMAttention[J].,2020,30(03):59.[doi:10. 3969 / j. issn. 1673-629X. 2020. 11. 011]
[4]尚福华,卢玉莹,曹茂俊.基于改进 LSTM 神经网络的测井曲线重构方法[J].计算机技术与发展,2022,32(06):198.[doi:10. 3969 / j. issn. 1673-629X. 2022. 06. 033]
　SHANG Fu-hua,LU Yu-ying,CAO Mao-jun.Well Logging Curve Reconstruction Method Based on Improved LSTM Neural Network[J].,2022,32(03):198.[doi:10. 3969 / j. issn. 1673-629X. 2022. 06. 033]
[5]吴春燕,李 理,黄鹏程,等.融合动态卷积注意力的机器阅读理解研究[J].计算机技术与发展,2023,33(07):160.[doi:10. 3969 / j. issn. 1673-629X. 2023. 07. 024]
　WU Chun-yan,LI Li,HUANG Peng-cheng,et al.Study on Machine Reading Comprehension Hybriding Dynamic Convolution Attention[J].,2023,33(03):160.[doi:10. 3969 / j. issn. 1673-629X. 2023. 07. 024]
[6]陈 刚.基于改进 GRU 的高速公路交通流量预测模型[J].计算机技术与发展,2023,33(07):208.[doi:10. 3969 / j. issn. 1673-629X. 2023. 07. 031]
　CHEN Gang.Highway Traffic Flow Prediction Model Based on Improved GRU[J].,2023,33(03):208.[doi:10. 3969 / j. issn. 1673-629X. 2023. 07. 031]

更新日期/Last Update: 2023-03-10