A Hybrid Deep Learning Model and Comparison for Wind Power Forecasting Considering Temporal-Spatial Feature Extraction
The inherent intermittency and uncertainty of wind power have brought challenges in accurate wind power output forecasting, which also cause tricky problems in the integration of wind power to the grid. In this paper, a hybrid deep learning model bidirectional long short term memory-convolutional ne...
Main Authors: | , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2020-11-01
|
Series: | Sustainability |
Subjects: | |
Online Access: | https://www.mdpi.com/2071-1050/12/22/9490 |
id |
doaj-2f2cadf80c8c4eafa0b718816c9f2bf3 |
---|---|
record_format |
Article |
spelling |
doaj-2f2cadf80c8c4eafa0b718816c9f2bf32020-11-25T04:04:28ZengMDPI AGSustainability2071-10502020-11-01129490949010.3390/su12229490A Hybrid Deep Learning Model and Comparison for Wind Power Forecasting Considering Temporal-Spatial Feature ExtractionHao Zhen0Dongxiao Niu1Min Yu2Keke Wang3Yi Liang4Xiaomin Xu5School of Economics and Management, North China Electric Power University, Beijing 102206, ChinaSchool of Economics and Management, North China Electric Power University, Beijing 102206, ChinaSchool of Economics and Management, North China Electric Power University, Beijing 102206, ChinaSchool of Economics and Management, North China Electric Power University, Beijing 102206, ChinaSchool of Management, Hebei Geo University, Shijiazhuang 050031, ChinaSchool of Economics and Management, North China Electric Power University, Beijing 102206, ChinaThe inherent intermittency and uncertainty of wind power have brought challenges in accurate wind power output forecasting, which also cause tricky problems in the integration of wind power to the grid. In this paper, a hybrid deep learning model bidirectional long short term memory-convolutional neural network (BiLSTM-CNN) is proposed for short-term wind power forecasting. First, the grey correlation analysis is utilized to select the inputs for forecasting model; Then, the proposed hybrid model extracts multi-dimension features of inputs to predict the wind power from the temporal-spatial perspective, where the Bi-LSTM model is utilized to mine the bidirectional temporal characteristics while the convolution and pooling operations of CNN are utilized to extract the spatial characteristics from multiple input time series. Lastly, a case study is conducted to verify the superiority of the proposed model. Other deep learning models (Bi-LSTM, LSTM, CNN, LSTM-CNN, CNN-BiLSTM, CNN-LSTM) are also simulated to conduct comparison from three aspects. The results show that the BiLSTM-CNN model has the best accuracy with the lowest RMSE of 2.5492, MSE of 6.4984, MAE of 1.7344 and highest R<sup>2</sup> of 0.9929. CNN has the fastest speed with an average computational time of 0.0741s. The hybrid model that mines the spatial feature based on the extracted temporal feature has a better performance than the model mines the temporal feature based on the extracted spatial feature.https://www.mdpi.com/2071-1050/12/22/9490wind powerdeep learningpower forecasthybrid modelBiLSTM-CNN |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Hao Zhen Dongxiao Niu Min Yu Keke Wang Yi Liang Xiaomin Xu |
spellingShingle |
Hao Zhen Dongxiao Niu Min Yu Keke Wang Yi Liang Xiaomin Xu A Hybrid Deep Learning Model and Comparison for Wind Power Forecasting Considering Temporal-Spatial Feature Extraction Sustainability wind power deep learning power forecast hybrid model BiLSTM-CNN |
author_facet |
Hao Zhen Dongxiao Niu Min Yu Keke Wang Yi Liang Xiaomin Xu |
author_sort |
Hao Zhen |
title |
A Hybrid Deep Learning Model and Comparison for Wind Power Forecasting Considering Temporal-Spatial Feature Extraction |
title_short |
A Hybrid Deep Learning Model and Comparison for Wind Power Forecasting Considering Temporal-Spatial Feature Extraction |
title_full |
A Hybrid Deep Learning Model and Comparison for Wind Power Forecasting Considering Temporal-Spatial Feature Extraction |
title_fullStr |
A Hybrid Deep Learning Model and Comparison for Wind Power Forecasting Considering Temporal-Spatial Feature Extraction |
title_full_unstemmed |
A Hybrid Deep Learning Model and Comparison for Wind Power Forecasting Considering Temporal-Spatial Feature Extraction |
title_sort |
hybrid deep learning model and comparison for wind power forecasting considering temporal-spatial feature extraction |
publisher |
MDPI AG |
series |
Sustainability |
issn |
2071-1050 |
publishDate |
2020-11-01 |
description |
The inherent intermittency and uncertainty of wind power have brought challenges in accurate wind power output forecasting, which also cause tricky problems in the integration of wind power to the grid. In this paper, a hybrid deep learning model bidirectional long short term memory-convolutional neural network (BiLSTM-CNN) is proposed for short-term wind power forecasting. First, the grey correlation analysis is utilized to select the inputs for forecasting model; Then, the proposed hybrid model extracts multi-dimension features of inputs to predict the wind power from the temporal-spatial perspective, where the Bi-LSTM model is utilized to mine the bidirectional temporal characteristics while the convolution and pooling operations of CNN are utilized to extract the spatial characteristics from multiple input time series. Lastly, a case study is conducted to verify the superiority of the proposed model. Other deep learning models (Bi-LSTM, LSTM, CNN, LSTM-CNN, CNN-BiLSTM, CNN-LSTM) are also simulated to conduct comparison from three aspects. The results show that the BiLSTM-CNN model has the best accuracy with the lowest RMSE of 2.5492, MSE of 6.4984, MAE of 1.7344 and highest R<sup>2</sup> of 0.9929. CNN has the fastest speed with an average computational time of 0.0741s. The hybrid model that mines the spatial feature based on the extracted temporal feature has a better performance than the model mines the temporal feature based on the extracted spatial feature. |
topic |
wind power deep learning power forecast hybrid model BiLSTM-CNN |
url |
https://www.mdpi.com/2071-1050/12/22/9490 |
work_keys_str_mv |
AT haozhen ahybriddeeplearningmodelandcomparisonforwindpowerforecastingconsideringtemporalspatialfeatureextraction AT dongxiaoniu ahybriddeeplearningmodelandcomparisonforwindpowerforecastingconsideringtemporalspatialfeatureextraction AT minyu ahybriddeeplearningmodelandcomparisonforwindpowerforecastingconsideringtemporalspatialfeatureextraction AT kekewang ahybriddeeplearningmodelandcomparisonforwindpowerforecastingconsideringtemporalspatialfeatureextraction AT yiliang ahybriddeeplearningmodelandcomparisonforwindpowerforecastingconsideringtemporalspatialfeatureextraction AT xiaominxu ahybriddeeplearningmodelandcomparisonforwindpowerforecastingconsideringtemporalspatialfeatureextraction AT haozhen hybriddeeplearningmodelandcomparisonforwindpowerforecastingconsideringtemporalspatialfeatureextraction AT dongxiaoniu hybriddeeplearningmodelandcomparisonforwindpowerforecastingconsideringtemporalspatialfeatureextraction AT minyu hybriddeeplearningmodelandcomparisonforwindpowerforecastingconsideringtemporalspatialfeatureextraction AT kekewang hybriddeeplearningmodelandcomparisonforwindpowerforecastingconsideringtemporalspatialfeatureextraction AT yiliang hybriddeeplearningmodelandcomparisonforwindpowerforecastingconsideringtemporalspatialfeatureextraction AT xiaominxu hybriddeeplearningmodelandcomparisonforwindpowerforecastingconsideringtemporalspatialfeatureextraction |
_version_ |
1724436625825464320 |