基于k-sums分段聚类的动态组合学习光伏短期功率预测

doi:10.16180/j.cnki.issn1007-7820.2024.04.010

摘要/Abstract

摘要：

目前单一模型预测精度存在难以随着功率波动保持最优的问题,为提高并网系统运行的稳定性和电网的节能调度,文中提出了一种基于k-sums分层聚类的动态学习组合光伏短期功率预测方法。利用k-sums算法经过分段聚类,将天气类型分为晴天A₁、多云A₂、阴雨天B。通过TCN(Temporal Convolutional Network)提取数据的时序特征,并结合GRU(Gate Recurrent Unit)建立融合提取时序特征模块的改进GRU结构,以达到对时序特征敏感的效果。将改进GRU结构与SVM(Support Vector Machine)动态组合,使用Elastic Net算法输出最佳权重值叠加得到最终预测值。文中采用江苏某地区的光伏发电功率数据及对应的气象数据对所提方法进行验证,结果表明动态组合学习模型的MAE(Mean Absolute Error)为1.888,RMSE(Root Mean Squared Error)为2.403。

关键词: k-sums, 分层聚类, TCN, 改进GRU, SVM, 动态组合学习, Elastic Net, 光伏短期功率预测

Abstract:

At present, the prediction accuracy of a single model is difficult to remain optimal with power fluctuation. To improve the stability of grid connected system operation and energy saving dispatching of power grid, this study proposes a dynamic learning combination short-term power prediction method based on k-sums hierarchical clustering. The weather types are divided into sunny day A₁, cloudy day A₂, and rainy day B through segmentation clustering using k-sums algorithm. The TCN (Temporal Convolutional Network) is used to extract the time series characteristics of data, and the GRU(Gate Recurrent Unit) structure of the fusion extraction time series characteristics module is established with GRU to achieve the effect of being sensitive to the time series characteristics. After dynamically combining the improved GRU structure with the SVM(Support Vector Machine), the Elastic Net algorithm is adopted to output the best weight value to obtain the final prediction value. The power data of photovoltaic power generation and corresponding meteorological data of a region in Jiangsu are used to verify the proposed method. The results show that the MAE(Mean Absolute Error) of the dynamic combination learning model is 1.888, and the RMSE(Root Mean Squared Error) is 2.403.

Key words: k-sums, hierarchical clustering, TCN, improve GRU, SVM, dynamic combination learning, Elastic Net, PV short-term power prediction

中图分类号:

TP183

吴家葆, 曾国辉, 张振华. 基于k-sums分段聚类的动态组合学习光伏短期功率预测[J]. 电子科技, 2024, 37(4): 69-76.

WU Jiabao, ZENG Guohui, ZHANG Zhenhua. Short-Term Photovoltaic Power Prediction Based on k-sums Quadratic Clustering and Dynamic Combination Learning[J]. Electronic Science and Technology, 2024, 37(4): 69-76.

图/表 9

图1

图2

图3

图4

表1

表2

表3

图5

表4

参考文献 20

[1]	国家能源局. 我国光伏发电并网装机容量突破3亿千瓦分布式发展成为新亮点[EB/OL].(2022-01-20) [2022-10-05] http://www.nea.gov.cn/2022-01/20/c_1310432517.htm
	National Energy Administration. The grid connected installed capacity of photovoltaic power generation in China has exceeded 300 million kilowatts, and distributed development has become a new highlight[EB/OL].(2022-01-20) [2022-10-05] http://www.nea.gov.cn/2022-01/20/c_1310432517.htm.
[2]	Eltawil M A, Zhao Z. Grid-connected photovoltaic power systems:Technical and potential problems-A review[J]. Renewable and Sustainable Energy Reviews, 2009, 14(1):112-129. doi: 10.1016/j.rser.2009.07.015
[3]	赖昌伟, 黎静华, 陈博, 等. 光伏发电出力预测技术研究综述[J]. 电工技术学报, 2019, 34(6):1201-1217.
	Lai Changwei, Li Jinghua, Chen Bo, et al. Review of photovoltaic power output prediction technology[J]. Transactions of China Electrotechnical Society, 2019, 34(6):1201-1217.
[4]	毛李帆, 姚建刚, 金永顺, 等. 中长期负荷预测的异常数据辨识与缺失数据处理[J]. 电网技术, 2010, 34(7):148-153.
	Mao Lifan, Yao Jiangang, Jin Yongshun, et al. Abnormal data identification and missing data filling in medium and long-term load forecasting[J]. Power System Technology, 2010, 34(7):148-153.
[5]	徐一伦, 张彬桥, 黄婧, 等. 考虑天气类型和相似日的IWPA-LSSVM光伏发电功率预测[J]. 中国电力, 2023, 56(2):143-149. doi: 10.11930/j.issn.1004-9649.202108059
	Xu Yilun, Zhang Binqiao, Huang Jing, et al. Forecast of photovoltaic power based on IWPA-LSSVM considering weather type and similar day[J]. Electric Power, 2023, 56(2):143-149. doi: 10.11930/j.issn.1004-9649.202108059
[6]	张雨金, 杨凌帆, 葛双冶, 等. 基于Kmeans-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 Yujin, Yang Lingfan, Ge Shuangye, et al. Short-term photovoltaic power forecasting based on Kmeans algorithm and support vector machine[J]. Power System Protection and Control, 2018, 46(21):118-124.
[7]	祁鑫, 王福忠, 张丽, 等. 基于SVD-LSTM的高校学生宿舍空调负荷预测[J]. 电子科技, 2020, 33(11):59-66.
	Qi Xin, Wang Fuzhong, Zhang Li, et al. Air conditioning load forecast of university student's dormitory based on SVD-LSTM[J]. Electronic Science and Technology, 2020, 33(11):59-66.
[8]	Gao X Y, Wang Y, Gao Y, et al. Short-term load forecasting model of GRU network based on deep learning framework[C]. Beijing: The Second IEEE Conference on Energy Internet and Energy System Integration, 2018:801-809.
[9]	黄圆, 魏云冰, 童东兵, 等. 基于VMD和改进TCN的短期光伏发电功率预测[J]. 电子科技, 2023, 36(3):42-49.
	Huang Yuan, Wei Yunbing, Tong Dongbing, et al. Short-term photovoltaic power prediction based on VMD andimproved TCN[J]. Electronic Science and Technology, 2023, 36(3):42-49.
[10]	邱明, 鲁冠军, 吴昊天, 等. 基于数据清洗与组合学习的光伏发电功率预测方法研究[J]. 可再生能源, 2020, 38(12):1583-1589.
	Qiu Ming, Lu Guanjun, Wu Haotian, et al. Short-term photovoltaic forecasting based on data cleansing and model combination[J]. Renewable Energy Resources, 2020, 38(12):1583-1589.
[11]	孟安波, 许炫淙, 陈嘉铭, 等. 基于强化学习和组合式深度学习模型的超短期光伏功率预测[J]. 电网技术, 2021, 45(12):4721-4728.
	Meng Anbo, Xu Xuancong, Chen Jiaming, et al. Ultra short term photovoltaic power prediction based on reinforcement learning and combined deep learning model[J]. Power System Technology, 2021, 45(12):4721-4728.
[12]	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 Rui, Yang Guohua, Zheng Haofeng, et al. Combined wind power prediction method based on CNN-LSTM&GRU with adaptive weights[J]. Electric Power, 2022, 55(5):47-56,110.
[13]	陈振宇, 刘金波, 李晨, 等. 基于LSTM与XGBoost组合模型的超短期电力负荷预测[J]. 电网技术, 2020, 44(2):614-620.
	Chen Zhenyu, Liu Jinbo, Li Chen, et al. Ultra short-term power load forecasting based on combined LSTM-XG-Boost model[J]. Power System Technology, 2020, 44(2):614-620.
[14]	Wan R Z, Mei S P, Wang J, et al. Multivariate temporal convolutional network:A deep neural networks approach form-ulti-variate time series forecasting[J]. Electronics, 2019, 8(8):876-879. doi: 10.3390/electronics8080876
[15]	He K M, Zhang X Y, Ren S Q, et al. Deep residual lear-ning for image recognition[C]Las Vegas: IEEE Conference on Computer Vision and Pattern Recognition, 2016:39-43.
[16]	邓带雨, 李坚, 张真源, 等. 基于EEMD-GRU-MLR的短期电力负荷预测[J]. 电网技术, 2020, 44(2):593-602.
	Deng Daiyu, Li Jian, Zhang Zhenyuan, et al. Short-term electric load forecasting based on EEMD-GRU-MLR[J]. Power System Technology, 2020, 44(2):593-602.
[17]	周恒俊, 王璇, 王志远, 等. 基于MIPCA与GRU网络的光伏出力短期预测方法[J]. 电力系统及其自动化学报, 2020, 32(9):55-62.
	Zhou Hengjun, Wang Xuan, Wang Zhiyuan, et al. Short-term photovoltaic output prediction method based on MIPCA and GRU network[J]. Proceedings of the CSU-EPSA, 2020, 32(9):55-62.
[18]	冯先丁, 魏镜弢, 吴张永, 等. 基于PCA-PSO-SVM的球磨机负荷预测研究[J]. 电子科技, 2022, 35(1):29-34.
	Feng Xianding, Wei Jingtao, Wu Zhangyong, et al. Research on load forecast of ball mill based on PCA-PSO- SVM[J]. Electronic Science and Technology, 2022, 35(1):29-34.
[19]	文彦飞, 王万雄. 基于FA-SVR-LSTM组合模型的短期电力负荷预测[J]. 电子科技, 2023, 36(9):1-7.
	Wen Yanfei, Wang Wanxiong. Short-term power load forecasting based on FA-SVR-LSTM combined model[J]. Electronic Science and Technology, 2023, 36(9):1-7.
[20]	宓登凯, 王彤, 相禹维, 等. 基于Elastic Net的暂态稳定裕度在线评估[J]. 电网技术, 2020, 44(1):19-26.
	Mi Dengkai, Wang Tong, Xiang Yuwei, et al. Elastic net based online assessment of power system transient stability margin[J]. Power System Technology, 2020, 44(1):19-26.

	晴天	多云	阴雨天
实际功率	1.000	1.000	1.000
总辐照度	0.953	0.937	0.892
法相直射辐照度	0.799	0.791	0.737
水平面散射辐照度	0.601	0.792	0.917
气温	0.194	0.258	0.213
气压	-0.031	-0.146	-0.198
相对湿度	-0.259	-0.211	-0.044

改进GRU	RMSE	MAE
2-1(64)	2.325	1.646
2-2(64,32)	2.112	1.478
2-3(64,32,16)	1.534	1.222
2-3(64,32,32)	1.797	1.740
3-3(64,32,16)	1.787	1.315
4-3(64,32,16)	1.996	1.483

天气	模型	RMSE	MAE
晴天	SVM	2.804	2.711
晴天	k-sums-SVM	2.427	2.376
多云	SVM	3.603	2.834
多云	k-sums-SVM	3.002	2.607
阴雨天	SVM	4.227	3.577
阴雨天	k-sums-SVM	3.988	3.113

	评价指标	GRU	SVM	改进 GRU	动态组合学习
晴天	RMSE	2.279	2.403	1.612	1.314
晴天	MAE	2.047	2.311	1.502	1.157
多云	RMSE	4.221	4.866	3.171	2.644
多云	MAE	3.713	4.153	2.247	1.746
阴雨天	RMSE	5.213	5.671	3.723	3.251
阴雨天	MAE	4.282	4.464	3.277	2.762
平均	RMSE	3.904	4.313	2.835	2.403
平均	MAE	3.347	3.643	2.342	1.888