Prediction of Streamflow Based on Dynamic Sliding Window LSTM
The streamflow of the upper reaches of the Yangtze River exhibits different timing and periodicity characteristics in different quarters and months of the year, which makes it difficult to predict. Existing sliding window-based methods usually use a fixed-size window, for which the window size selec...
Main Authors: | , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2020-10-01
|
Series: | Water |
Subjects: | |
Online Access: | https://www.mdpi.com/2073-4441/12/11/3032 |
id |
doaj-748665fea26e4365be8e4e90a5ef1939 |
---|---|
record_format |
Article |
spelling |
doaj-748665fea26e4365be8e4e90a5ef19392020-11-25T02:42:08ZengMDPI AGWater2073-44412020-10-01123032303210.3390/w12113032Prediction of Streamflow Based on Dynamic Sliding Window LSTMLimei Dong0Desheng Fang1Xi Wang2Wei Wei3Robertas Damaševičius4Rafał Scherer5Marcin Woźniak6Upper Changjiang River Bureau of Hydrological and Water Resources Survey, Changjiang Water Resources Commission, Chongqing 400020, ChinaUpper Changjiang River Bureau of Hydrological and Water Resources Survey, Changjiang Water Resources Commission, Chongqing 400020, ChinaUpper Changjiang River Bureau of Hydrological and Water Resources Survey, Changjiang Water Resources Commission, Chongqing 400020, ChinaSchool of Computer Science and Engineering, Xi’an University of Technology, Xi’an 710048, ChinaDepartment of Applied Informatics, Vytautas Magnus University, 44404 Kaunas, LithuaniaDepartment of Intelligent Computer Systems, Częstochowa University of Technology, 42200 Częstochowa, PolandFaculty of Applied Mathematics, Silesian University of Technology, 44100 Gliwice, PolandThe streamflow of the upper reaches of the Yangtze River exhibits different timing and periodicity characteristics in different quarters and months of the year, which makes it difficult to predict. Existing sliding window-based methods usually use a fixed-size window, for which the window size selection is random, resulting in large errors. This paper proposes a dynamic sliding window method that reflects the different timing and periodicity characteristics of the streamflow in different months of the year. Multiple datasets of different months are generated using a dynamic window at first, then the long-short term memory (LSTM) is used to select the optimal window, and finally, the dataset of the optimal window size is used for verification. The proposed method was tested using the hydrological data of Zhutuo Hydrological Station (China). A comparison between the flow prediction data and the measured data shows that the prediction method based on a dynamic sliding window LSTM is more accurate by 8.63% and 3.85% than the prediction method based on fixed window LSTM and the dynamic sliding window back-propagation neural network, respectively. This method can be generally used for the time series data prediction with different periodic characteristics.https://www.mdpi.com/2073-4441/12/11/3032streamflowflow predictiondynamic sliding windowdeep learningneural networkLSTM |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Limei Dong Desheng Fang Xi Wang Wei Wei Robertas Damaševičius Rafał Scherer Marcin Woźniak |
spellingShingle |
Limei Dong Desheng Fang Xi Wang Wei Wei Robertas Damaševičius Rafał Scherer Marcin Woźniak Prediction of Streamflow Based on Dynamic Sliding Window LSTM Water streamflow flow prediction dynamic sliding window deep learning neural network LSTM |
author_facet |
Limei Dong Desheng Fang Xi Wang Wei Wei Robertas Damaševičius Rafał Scherer Marcin Woźniak |
author_sort |
Limei Dong |
title |
Prediction of Streamflow Based on Dynamic Sliding Window LSTM |
title_short |
Prediction of Streamflow Based on Dynamic Sliding Window LSTM |
title_full |
Prediction of Streamflow Based on Dynamic Sliding Window LSTM |
title_fullStr |
Prediction of Streamflow Based on Dynamic Sliding Window LSTM |
title_full_unstemmed |
Prediction of Streamflow Based on Dynamic Sliding Window LSTM |
title_sort |
prediction of streamflow based on dynamic sliding window lstm |
publisher |
MDPI AG |
series |
Water |
issn |
2073-4441 |
publishDate |
2020-10-01 |
description |
The streamflow of the upper reaches of the Yangtze River exhibits different timing and periodicity characteristics in different quarters and months of the year, which makes it difficult to predict. Existing sliding window-based methods usually use a fixed-size window, for which the window size selection is random, resulting in large errors. This paper proposes a dynamic sliding window method that reflects the different timing and periodicity characteristics of the streamflow in different months of the year. Multiple datasets of different months are generated using a dynamic window at first, then the long-short term memory (LSTM) is used to select the optimal window, and finally, the dataset of the optimal window size is used for verification. The proposed method was tested using the hydrological data of Zhutuo Hydrological Station (China). A comparison between the flow prediction data and the measured data shows that the prediction method based on a dynamic sliding window LSTM is more accurate by 8.63% and 3.85% than the prediction method based on fixed window LSTM and the dynamic sliding window back-propagation neural network, respectively. This method can be generally used for the time series data prediction with different periodic characteristics. |
topic |
streamflow flow prediction dynamic sliding window deep learning neural network LSTM |
url |
https://www.mdpi.com/2073-4441/12/11/3032 |
work_keys_str_mv |
AT limeidong predictionofstreamflowbasedondynamicslidingwindowlstm AT deshengfang predictionofstreamflowbasedondynamicslidingwindowlstm AT xiwang predictionofstreamflowbasedondynamicslidingwindowlstm AT weiwei predictionofstreamflowbasedondynamicslidingwindowlstm AT robertasdamasevicius predictionofstreamflowbasedondynamicslidingwindowlstm AT rafałscherer predictionofstreamflowbasedondynamicslidingwindowlstm AT marcinwozniak predictionofstreamflowbasedondynamicslidingwindowlstm |
_version_ |
1724775133302751232 |