为保障电网供需平衡和安全稳定运行,提高大型光伏电站功率预测的精度,提出一种基于K中心点聚类算法（K-medoids）、梯度提升树（GBDT）和粒子群优化算法（PSO）组合优化的长短期记忆神经网络（LSTM）的光伏功率短期预测模型。首先,采用K-medoids聚类算法对大规模光伏发电数据样本中的天气数据进行不同类别聚类,分为晴天、阴天和雨/雪天3种天气类型;然后,在已有数据基础上构造特征工程,使用GBDT算法分别进行特征重要性分析,筛选出对光伏功率预测具有显著影响的特征,并构建合适大小结构的优化数据集;最后,将重构后的数据集代入PSO算法优化的LSTM模型进行训练,以建立短期预测模型。实验结果表明,该模型拥有更高预测精度,相比单一LSTM模型,在雨/雪天下的RMSE指标降低了12.19%。
Abstract
In this work, a novel approach for predicting photovoltaic power is proposed, which aims to ensuring the stability of the power grid and maintaining a balance between supply and demand. The proposed model integrates the K-medoids clustering algorithm, gradient boosting decision trees （GBDT）, particle swarm optimization （PSO）, and long short-term memory （LSTM） neural networks. Firstly, the K-medoids clustering algorithm is utilized to classify weather data from a large-scale photovoltaic power generation dataset into three different weather types： sunny, cloudy, and rainy/snowy. Afterwards, the feature engineering approach is employed to the extension of existing dataset, and GBDT is employed to analyze the importance of various features, thereby identifying the significant factors influencing photovoltaic power prediction, and further construct an optimized dataset with suitable size. Finally, the reconstructed dataset is further used to train an LSTM model, which is optimized using the PSO algorithm, thus establishing an accurate short-term prediction model. Experimental findings demonstrate that the proposed model achieves a higher prediction accuracy. Moreover, when compared to a single LSTM model, the RMSE indicator is reduced by 12.19% specifically under rainy weather conditions.
关键词
光伏发电 /
功率预测 /
机器学习 /
长短期记忆网络 /
优化算法 /
粒子群算法
Key words
photovoltaic power /
power forecasting /
machine learning /
long short-term memory /
optimization algorithms /
particle swarm algorithm
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 国家能源局发布1~7月份全国电力工业统计数据[N]. 中国电力报, 2023.
The National Energy Administration released statistical data on the national power industry from January to July[N].The National Energy Administration released statistical data on the national power industry from January to July[N]. China electric power news, 2023.
[2] 王飞, 米增强, 杨奇逊, 等. 基于神经网络与关联数据的光伏电站发电功率预测方法[J]. 太阳能学报, 2012, 33(7): 1171-1177.
WANG F, MI Z Q, YANG Q X, et al.Power forecasting approach of pv plant based on ann and relevant data[J]. Acta energiae solaris sinica, 2012, 33(7): 1171-1177.
[3] 耿博, 高贞彦, 白恒远, 等. 结合相似日GA-BP神经网络的光伏发电预测[J]. 电力系统及其自动化学报, 2017, 29(6): 118-123.
GENG B, GAO Z Y, BAI H Y, et al.PV generation forecasting combined with similar days and GA-BP neural network[J]. Proceedings of the CSU-EPSA, 2017, 29(6): 118-123.
[4] LIU L Y, LIU D R, SUN Q, et al.Forecasting power output of photovoltaic system using A BP network method[J]. Energy procedia, 2017, 142: 780-786.
[5] 张佳伟, 张自嘉. 基于PSO-BP神经网络的短期光伏系统发电预测[J]. 可再生能源, 2012, 30(8): 28-32.
ZHANG J W, ZHANG Z J.Short-term photovoltaic system power forecasting based on PSO-BP neural network[J]. Renewable energy resources, 2012, 30(8): 28-32.
[6] 李争, 罗晓瑞, 张杰, 等. 基于改进麻雀搜索算法的光伏功率短期预测[J]. 太阳能学报, 2023, 44(6): 284-289.
LI Z, LUO X R, ZHANG J, et al.Short term prediction of photovoltaic power based on improved sparrow search algorithm[J]. Acta energiae solaris sinica, 2023, 44(6): 284-289.
[7] 王育飞, 付玉超, 孙路, 等. 基于混沌-RBF神经网络的光伏发电功率超短期预测模型[J]. 电网技术, 2018, 42(4): 1110-1116.
WANG Y F, FU Y C, SUN L, et al.Ultra-short term prediction model of photovoltaic output power based on chaos-RBF neural network[J]. Power system technology, 2018, 42(4): 1110-1116.
[8] 薛阳, 燕宇铖, 贾巍, 等. 基于改进灰狼算法优化长短期记忆网络的光伏功率预测[J]. 太阳能学报, 2023, 44(7): 207-213.
XUE Y, YAN Y C, JIA W, et al.Photovoltaic power prediction model based on IGWO-LSTM[J]. Acta energiae solaris sinica, 2023, 44(7): 207-213.
[9] 周浩杰, 杨建卫, 王尊, 等. 基于LSTM光伏发电功率超短期预测模型研究[J]. 电源技术, 2023, 47(6): 785-789.
ZHOU H J, YANG J W, WANG Z, et al.Research on ultra-short term prediction model of photovoltaic power generation based on LSTM[J]. Chinese journal of power sources, 2023, 47(6): 785-789.
[10] 张雨金, 杨凌帆, 葛双冶, 等. 基于K-means-SVM的短期光伏发电功率预测[J]. 电力系统保护与控制, 2018, 46(21): 118-124.
ZHANG Y J, YANG L F, GE S Y, et al.Short-term photovoltaic power forecasting based on K-means algorithm and support vector machine[J]. Power system protection and control, 2018, 46(21): 118-124.
[11] 孟安波, 陈嘉铭, 黎湛联, 等. 基于相似日理论和CSO-WGPR的短期光伏发电功率预测[J]. 高电压技术, 2021, 47(4): 1176-1184.
MENG A B, CHEN J M, LI Z L, et al.Short-term photovoltaic power generation prediction based on similar day theory and CSO-WGPR[J]. High voltage engineering, 2021, 47(4): 1176-1184.
[12] LI Q, ZHANG X Y, MA T J, et al.A multi-step ahead photovoltaic power forecasting model based on TimeGAN, Soft DTW-based K-medoids clustering, and a CNN-GRU hybrid neural network[J]. Energy reports, 2022, 8: 10346-10362.
[13] 王春平. 基于光伏功率预测的分布式能源系统优化[D]. 上海: 上海交通大学, 2020.
WANG C P.Optimization of distributed energy system based on photovoltaic power prediction[D]. Shanghai: Shanghai Jiao Tong University, 2020.
[14] 何泊延. 基于GBDT的数据分类方法的研究及其应用[D]. 吉林: 东北电力大学, 2021.
HE B Y.Research and application of data classification method based on GBDT[D]. Jilin: Northeast Dianli University, 2021.
[15] KENNEDY J, EBERHART R.Particle swarm optimization[C]//Proceedings of ICNN'95-International Conference on Neural Networks. Perth, WA, Australia, 1995: 1942-1948.
[16] SHERSTINSKY A.Fundamentals of recurrent neural network (RNN) and long short-term memory(LSTM) network[J]. Physica D: nonlinear phenomena, 2020, 404: 132306.
[17] 卢捷, 于立军, 郑培, 等. 风氢耦合系统超前控制策略研究[J]. 太阳能学报, 2022, 43(3): 53-60.
LU J, YU L J, ZHENG P, et al.Research on advanced control strategy of wind hydrogen coupling system[J]. Acta energiae solaris sinica, 2022, 43(3): 53-60.
[18] 国家市场监督管理总局, 国家标准化管理委员会. 调度侧风电或光伏功率预测系统技术要求: GB/T 40607—2021[S]. 北京: 中国标准出版社, 2021.
State Administration for Market Regulation, Standardization Administration of the People's Republic of China. Technical requirements for dispatching side forecasting system of wind or photovoltaic power: GB/T 40607—2021[S]. Beijing: Standards Press of China, 2021.
基金
上海交通大学-国家电投“未来能源计划联合基金”（202110）
PDF(2335 KB)
