干旱区生态环境知识资源中心(Arid): Forecasting the potential of reclaimed water using signal decomposition and deep learning

Arid

DOI	10.1016/j.jwpe.2024.105770
	Forecasting the potential of reclaimed water using signal decomposition and deep learning
	Chen, Yinglong; Zhang, Hongling; Peng, Jingkai; Ma, Shilong; Xu, Tengsheng; Tang, Lian
通讯作者	Tang, L
来源期刊	JOURNAL OF WATER PROCESS ENGINEERING
ISSN	2214-7144
出版年	2024
卷号	65
英文摘要	In the context of increasingly strained global water resources, the reuse of reclaimed water holds significant economic and social benefits. Accurate forecasting of reclaimed water potential is crucial for precise allocation and optimization studies of reclaimed water utilization. Conventional reclaimed water potential forecasting models, limited to annual predictions, fail to accurately reflect reclaimed water volume variations within different periods, particularly in arid and semi-arid areas with significant seasonal fluctuations, thus failing to meet the needs of detailed water resource management. This study analyzes the seasonal trends in reclaimed water potential and meteorological impacts on reclaimed water potential. The analysis reveals an increase in reclaimed water volume in spring, peaking in summer, decreasing in autumn, and reaching the lowest levels in winter, with positive correlations to temperature and rainfall, and a slight negative correlation to atmospheric pressure. Additionally, this study proposes a hybrid forecasting model utilizing signal decomposition and deep learning. The model improves prediction accuracy and scale by optimizing data input with signal decomposition algorithms and leveraging the temporal processing capabilities of long short-term memory neural network (LSTM) with the feature extraction of Convolutional Neural Network (CNN). The model achieves R2 values of 0.96 and 0.92, with MAPE values of 3.38 % and 2.68 %, respectively. The model's superior predictive accuracy, generality, and robustness have been validated across various scenarios.
英文关键词	Complete ensemble empirical mode decompositionwith adaptivenoise(CEEMDAN) Deep learning Long short-term memory neural network (LSTM) Convolutional neural network (CNN) Reclaimed water volumes forecasting
类型	Article
语种	英语
收录类别	SCI-E
WOS记录号	WOS:001279148800001
WOS关键词	SIMULATION ; REUSE
WOS类目	Engineering, Environmental ; Engineering, Chemical ; Water Resources
WOS研究方向	Engineering ; Water Resources
资源类型	期刊论文
条目标识符	http://119.78.100.177/qdio/handle/2XILL650/404762
推荐引用方式 GB/T 7714	Chen, Yinglong,Zhang, Hongling,Peng, Jingkai,et al. Forecasting the potential of reclaimed water using signal decomposition and deep learning[J],2024,65.
APA	Chen, Yinglong,Zhang, Hongling,Peng, Jingkai,Ma, Shilong,Xu, Tengsheng,&Tang, Lian.(2024).Forecasting the potential of reclaimed water using signal decomposition and deep learning.JOURNAL OF WATER PROCESS ENGINEERING,65.
MLA	Chen, Yinglong,et al."Forecasting the potential of reclaimed water using signal decomposition and deep learning".JOURNAL OF WATER PROCESS ENGINEERING 65(2024).