基于增量学习的CNN-LSTM光伏功率预测

Abstract
Figure/Table
References
Related Citation (3)

Download: PDF (1658 KB) HTML (1 KB)
Export: BibTeX | EndNote (RIS)

Abstract Most photovoltaic (PV) power prediction models adopt batch offline training, which poses a challenge on dealing with limited training data for newly established PV power plants. In order to address this issue, a PV power prediction model based on a combination of convolutional neural network (CNN) and long short-term memory (LSTM) network using incremental learning is proposed. Firstly, the CNN is used to extract the features of the meteorological data, and the power prediction is carried out through the LSTM network. The CNN-LSTM hybrid model is used for background learning, to train a baseline model that can be used for incremental learning. Secondly, incremental learning training is carried out according to different time spans to realize the online update of the model. In order to solve the problem of catastrophic forgetting in incremental learning, this paper uses the elastic weight consolidation (EWC) algorithm and the online elastic consolidation (Online_EWC) algorithm. Experimental results show that, compared with unconstrained incremental learning, incremental learning using EWC and Online_EWC methods can significantly alleviate the problem of catastrophic forgetting and reduce the prediction mean absolute error (MAE) and root mean square error (RMSE), up to 21.7% and 18.3%, respectively. At the same time, the time cost of incremental learning is significantly lower than that of traditional batch learning.

Key words： photovoltaic power prediction long short-term memory (LSTM) networks incremental learning elastic weight consolidation (EWC) algorithm

Received: 30 October 2023

	Service

	E-mail this article
	Add to my bookshelf
	Add to citation manager
	E-mail Alert
	RSS
	Articles by authors
	YAN Luhan
	LIN Peijie
	CHENG Shuying
	CHEN Zhicong
	LU Xiaoyang

Cite this article:

YAN Luhan,LIN Peijie,CHENG Shuying等. CNN-LSTM photovoltaic power prediction based on incremental learning[J]. Electrical Engineering, 2024, 25(5): 31-40.

URL:

http://dqjs.cesmedia.cn/EN/Y2024/V25/I5/31

[1] 王梦犀. 太阳能光伏技术在国家公园中的应用分析[J]. 电气技术, 2020, 21(7): 112-115, 124.
[2] 谭大帅, 戴彬, 郭刚, 等. 分布式光伏管控平台的设计与实现[J]. 电气技术, 2023, 24(2): 41-51.
[3] 涂彦昭, 高伟, 杨耿杰. 一种基于卷积神经网络和长短期记忆网络的光伏系统故障辨识方法[J]. 电气技术, 2022, 23(2): 48-54.
[4] 陈振祥, 林培杰, 程树英, 等. 基于K-means++和混合深度学习的光伏功率预测[J]. 电气技术, 2021, 22(9): 7-13, 33.
[5] 唐斯, 陈新楚, 郑松. 基于注意力与多尺度卷积神经网络的电机轴承故障诊断[J]. 电气技术, 2020, 21(11): 32-38.
[6] LUO Xing, ZHANG Dongxiao, ZHU Xu.Deep learning based forecasting of photovoltaic power generation by incorporating domain knowledge[J]. Energy, 2021, 225: 120240.
[7] 江晶晶, 窦真兰, 杨海涛, 等. 贫乏信息下基于深度迁移学习的智慧建筑负荷预测方法[J]. 电气技术, 2022, 23(5): 55-61, 72.
[8] COPPOCK H W, FREUND J E.All-or-none versus incremental learning of errorless shock escapes by the rat[J]. Science, 1962, 135(3500): 318-319.
[9] 范兴明, 王超, 张鑫, 等. 基于增量学习相关向量机的锂离子电池SOC预测方法[J]. 电工技术学报, 2019, 34(13): 2700-2708.
[10] 尚鑫. 基于增量学习的智慧化城镇供水量预测算法研究[D]. 重庆: 重庆大学, 2020.
[11] SARMAS E, STROMPOLAS S, MARINAKIS V, et al.An incremental learning framework for photovoltaic production and load forecasting in energy micro- grids[J]. Electronics, 2022, 11(23): 3962.
[12] RATCLIFF R.Connectionist models of recognition memory: constraints imposed by learning and forgetting functions[J]. Psychological Review, 1990, 97(2): 285-308.
[13] KIRKPATRICK J, PASCANU R, RABINOWITZ N, et al.Overcoming catastrophic forgetting in neural networks[J]. Proceedings of the National Academy of Sciences of the United States of America, 2017, 114(13): 3521-3526.
[14] 张斌, 周奕涛. 基于EWC算法的DDoS攻击检测模型参数更新方法[J]. 电子与信息学报, 2021, 43(10): 2928-2935.
[15] 张翼, 朱永利. 结合知识蒸馏和图神经网络的局部放电增量识别方法[J]. 电工技术学报, 2023, 38(5): 1390-1400.
[16] 潘锦业, 王苗苗, 阚威, 等. 基于Adam优化算法和长短期记忆神经网络的锂离子电池荷电状态估计方法[J]. 电气技术, 2022, 23(4): 25-30, 36.
[17] 许晓晖, 郝春昊, 徐成虎, 等. 基于布谷鸟搜索—反向传播神经网络的弥散供氧PID控制方法[J]. 电气技术, 2023, 24(8): 12-21.
[18] 王艺博. 基于莱维飞行蜉蝣优化算法的光伏阵列最大功率点跟踪研究[J]. 电气技术, 2022, 23(1): 64-69.
[19] BIFET A, GAVALDA R, HOLMES G, et al.Machine learning for data streams: with practical examples in MOA[M]. Cambridge: MIT Press, 2018.
[20] SCHWARZ J, LUKETINA J, CZARNECKI W M, et al. Progress & compress: a scalable framework for continual learning[EB/OL]. 2018: arXiv: 1805.06370. http://arxiv.org/abs/1805.06370.pdf.