A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Offshore wind speed forecasting at different heights by using ensemble empirical mode decomposition and deep learning models
Abstract Comparison of six different variants of deep learning based models namely Convolutional Neural Network (CNN), Stacked Long Short-Term Memory (LSTM), Bidirectional LSTM, CNN-LSTM, Multilayer Perceptron, and Two-dimensional Convolutional LSTM based forecasting models is done in this paper for offshore wind speed forecasting. Models are developed and trained for one-step ahead short-term offshore wind speed forecasting. Ensemble Empirical Mode Decomposition (EEMD) of wind speed data is also done. All models are verified at two different offshore sites. At site-1 located near Gulf of Khambat (India), four experiments are performed by using offshore wind speed data of four heights (180, 160, 140, and 120 m) that are measured by a Light Detection and Ranging (LiDAR) installed at 25 km off the coast of Gujarat state in the Arabian Sea. At site-2 located near Gulf of Mannar at Laccadive Sea in the Indian Ocean, two experiments are performed by using wind speed data of two heights (102 and 80 m) that are measured by a seashore meteorological mast located at Dhanushkodi in Tamil Nadu state of India. Experimental results confirm that EEMD reduces the forecasting error. The superiority of deep learning model is site specific and changes with change in site.
Offshore wind speed forecasting at different heights by using ensemble empirical mode decomposition and deep learning models
Abstract Comparison of six different variants of deep learning based models namely Convolutional Neural Network (CNN), Stacked Long Short-Term Memory (LSTM), Bidirectional LSTM, CNN-LSTM, Multilayer Perceptron, and Two-dimensional Convolutional LSTM based forecasting models is done in this paper for offshore wind speed forecasting. Models are developed and trained for one-step ahead short-term offshore wind speed forecasting. Ensemble Empirical Mode Decomposition (EEMD) of wind speed data is also done. All models are verified at two different offshore sites. At site-1 located near Gulf of Khambat (India), four experiments are performed by using offshore wind speed data of four heights (180, 160, 140, and 120 m) that are measured by a Light Detection and Ranging (LiDAR) installed at 25 km off the coast of Gujarat state in the Arabian Sea. At site-2 located near Gulf of Mannar at Laccadive Sea in the Indian Ocean, two experiments are performed by using wind speed data of two heights (102 and 80 m) that are measured by a seashore meteorological mast located at Dhanushkodi in Tamil Nadu state of India. Experimental results confirm that EEMD reduces the forecasting error. The superiority of deep learning model is site specific and changes with change in site.
Offshore wind speed forecasting at different heights by using ensemble empirical mode decomposition and deep learning models
Saxena, Bharat Kumar (author) / Mishra, Sanjeev (author) / Rao, Komaragiri Venkata Subba (author)
Applied Ocean Research ; 117
2021-10-21
Article (Journal)
Electronic Resource
English
Characterizing Nonstationary Wind Speed Using Empirical Mode Decomposition
| British Library Online Contents | 2004
Characterizing Nonstationary Wind Speed Using Empirical Mode Decomposition
| Online Contents | 2004
Leakage Detection in Pipelines Using Ensemble Empirical Mode Decomposition
| British Library Conference Proceedings | 2010
| American Institute of Physics | 2016