Forecasting wind speed in Zabol city: a comparative study of CNN, LSTM, and CNN-LSTM models

Bagherpoor, Tohid; Kazemi Sormoli, Somayeh

doi:10.22034/jelsa.2024.416818.1052

Forecasting wind speed in Zabol city: a comparative study of CNN, LSTM, and CNN-LSTM models

Document Type : Original research article

Authors

Tohid Bagherpoor ¹

Somayeh Kazemi Sormoli ²

¹ Department of Agronomy, Faculty of Agriculture, University of Zabol, Zabol, Iran

² Department of Geo-information Engineering, School of Computer Science, China University of Geoscience (Wuhan), Wuhan, China

10.22034/jelsa.2024.416818.1052

Abstract

One of the most essential concerns in renewable energy planning, weather forecasting, and environmental research is accurately predicting wind speed. This study evaluates and compares three machine learning methods: Long Short-Term Memory (LSTM), Convolutional Neural Network (CNN), and a hybrid CNN-LSTM. The dataset includes daily-averaged wind speed (m/s), the day of the year, and additional meteorological variables recorded at Zabol station from 2010 to 2021. The dataset was preprocessed using Min-Max normalization and then split into four seasonal subsets: spring, summer, autumn, and winter. Each season was further divided into distinct subsets for training, validation, and testing in a 7:2:1 ratio. Python 3.8 was utilized for model development and data preparation using TensorFlow and Keras libraries. Performance evaluation metrics included Pearson correlation coefficient, root mean square error (RMSE), mean absolute error (MAE), and mean directional absolute percentage error (MDAPE). The results highlight that the CNN-LSTM hybrid model achieves superior accuracy compared to the standalone CNN and LSTM models. CNN effectively captures spatial patterns, while LSTM manages long-term temporal dependencies. The integration of these strengths in the CNN-LSTM model enhances forecasting accuracy across diverse conditions. Notably, the CNN-LSTM model achieved the highest accuracy in autumn, with an R² value of 0.980, showcasing its robustness in capturing wind speed variations across diverse seasonal conditions. This research highlights the strengths and limitations of each model. The main objective of this study is to precisely predict wind speed to optimize energy production, improve weather forecasting, and evaluate environmental impacts. Future research could explore additional meteorological variables, hybrid modeling, improved preprocessing techniques, and the use of larger datasets.

Highlights

A novel integrated model was introduced for wind speed prediction.
The proposed model demonstrated superior performance compared to three benchmark models across four seasons.
The research highlighted the timeliness and significance by pointing out the absence of comprehensive studies comparing CNN, LSTM, and CNN-LSTM models in wind speed prediction.

Keywords

Convolutional neural network

Hybrid models

Long short-term memory

Time series analysis

Wind speed prediction

Ali, M. H. (2012). Wind energy systems: Solutions for power quality and stabilization: Crc Press.

Bali, V., Kumar, A., & Gangwar, S. (2019). Deep learning based wind speed forecasting- a review. Paper presented at the 9th International Conference on Cloud Computing, Data Science & Engineering (Confluence), Noida, India. doi: 10.1109/confluence.2019.8776923

Bastos, B. Q., Oliveira, F. L. C., & Milidiu, R. L. (2021). U-convolutional model for spatio-temporal wind speed forecasting. International Journal of Forecasting, 37(2), 949-970. doi: 10.1016/j.ijforecast.2020.10.007

Chen, Y., Wang, Y., Dong, Z., Su, J., Han, Z., Zhou, D., Zhao, Y., & Bao, Y. (2021). 2-d regional short-term wind speed forecast based on cnn-lstm deep learning model. Energy Conversion and Management, 244, 114451. doi: 10.1016/j.enconman.2021.114451

Dauphin, Y. N., Fan, A., Auli, M., & Grangier, D. (2017). Language modeling with gated convolutional networks. Paper presented at the the 34th International Conference on Machine Learning, Sydney, Australia.

Frazier, P. I., & Wang, J. (2016). Bayesian optimization for materials design. Information science for materials discovery and design, 45-75. doi: 10.1007/978-3-319-23871-5_3

Hinton, G. E., & Salakhutdinov, R. R. (2006). Reducing the dimensionality of data with neural networks. SCIENCE, 313(5786), 504-507. doi: 10.1126/science.1127647

Hochreiter, S., & Schmidhuber, J. u. (1997). Long short-term memory. Neural Computation 9(8), 1735-1780.

doi: 10.1162/neco.1997.9.8.1735

Hou, J., Wang, Y., Zhou, J., & Tian, Q. (2022). Prediction of hourly air temperature based on cnn–lstm, geomatics. Natural Hazards and Risk, 13(1), 1962-1986. doi: 10.1080/19475705.2022.2102942

Jaseena, K. U., & Kovoor, B. C. (2021). Decomposition-based hybrid wind speed forecasting model using deep bidirectional lstm networks. Energy Conversion and Management, 234. doi: 10.1016/j.enconman.2021.113944

Karthik, R., Hariharan, M., Anand, S., Mathikshara, P., Johnson, A., & Menaka, R. (2020). Attention embedded residual cnn for disease detection in tomato leaves. Applied Soft Computing, 86, 105933. doi: 10.1016/j.asoc.2019.105933

Kusiak, A., Zhang, Z., & Verma, A. (2013). Prediction, operations, and condition monitoring in wind energy. Energy, 60, 1-12. doi: 10.1016/j.energy.2013.07.051

Lecun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998). Gradient-based learning applied to document recognition. Paper presented at the in Proceedings of the IEEE. doi: 10.1109/5.726791

Liu, H., Duan, Z., Wu, H., Li, Y., & Dong, S. (2019). Wind speed forecasting models based on data decomposition, feature selection and group method of data handling network. Measurement, 148, 106971. doi: 10.1016/j.measurement.2019.106971.

Liu, H., Yu, C., Wu, H., Duan, Z., & Yan, G. (2020). A new hybrid ensemble deep reinforcement learning model for wind speed short term forecasting. Energy, 202, 117794. doi: 10.1016/j.energy.2020.117794

Lv, S.-X., & Wang, L. (2022). Deep learning combined wind speed forecasting with hybrid time series decomposition and multi-objective parameter optimization. Applied Energy, 311, 118674. doi: 10.1016/j.apenergy.2022.118674

Memarzadeh, G., & Keynia, F. (2020). A new short-term wind speed forecasting method based on fine-tuned lstm neural network and optimal input sets. Energy Conversion and Management, 213, 112824. doi: 10.1016/j.enconman.2020.112824

Neshat, M., Nezhad, M. M., Abbasnejad, E., Mirjalili, S., Groppi, D., Heydari, A., . . . Wagner, M. (2021a). Wind turbine power output prediction using a new hybrid neuro-evolutionary method. Energy, 229. doi: 10.1016/j.energy.2021.120617

Neshat, M., Nezhad, M. M., Abbasnejad, E., Mirjalili, S., Tjernberg, L. B., Astiaso Garcia, D., Alexander, B., & Wagner, M. (2021b). A deep learning-based evolutionary model for short-term wind speed forecasting: A case study of the lillgrund offshore wind farm. Energy Conversion and Management, 236, 114002. doi: 10.1016/j.enconman.2021.114002

Neshat, M., Nezhad, M. M., Abbasnejad, E., Tjernberg, L. B., Garcia, D. A., Alexander, B., & Wagner, M. (2020). An evolutionary deep learning method for short-term wind speed prediction: A case study of the lillgrund offshore wind farm. ArXiv, abs/2002.09106. doi: 10.48550/arxiv.2002.09106

Qiao, D., Wu, S., Li, G., You, J., Zhang, J., & Shen, B. (2022). Wind speed forecasting using multi-site collaborative deep learning for complex terrain application in valleys. Renewable Energy, 189, 231-244. doi: 10.1016/j.renene.2022.02.095

Radhika, Y., & Shashi, M. (2009). Atmospheric temperature prediction using support vector machines. International Journal of Computer Theory and Engineering, 55-58. doi: 10.7763/ijcte.2009.V1.9

Schepers, J. G., & Snel, H. (2007). Model experiments in controlled conditions. ECN Report: ECN-E-07-042, 484.

Shobana Devi, A., Maragatham, G., Boopathi, K., Lavanya, M. C., & Saranya, R. (2021). Long-term wind speed forecasting—a review. Paper presented at the Artificial Intelligence Techniques for Advanced Computing Applications, Singapore. doi: 10.1007/978-981-15-5329-5_9

Sun, W., Liu, M., & Liang, Y. (2015). Wind speed forecasting based on feemd and lssvm optimized by the bat algorithm. Energies, 8(7), 6585-6607. doi: 10.3390/en8076585

TuTiempo.Net. (2010-2021). Global climate data, from https://en.tutiempo.net/climate

Vanderwende, B., & Lundquist, J. K. (2015). Could crop height affect the wind resource at agriculturally productive wind farm sites? Boundary-Layer Meteorology, 158(3), 409-428. doi: 10.1007/s10546-015-0102-0

Yang, R., Liu, H., Nikitas, N., Duan, Z., Li, Y., & Li, Y. (2022). Short-term wind speed forecasting using deep reinforcement learning with improved multiple error correction approach. Energy, 239, 122128. doi: 10.1016/j.energy.2021.122128

Yildiz, C., Acikgoz, H., Korkmaz, D., & Budak, U. (2021). An improved residual-based convolutional neural network for very short-term wind power forecasting. Energy Conversion and Management, 228, 113731. doi: 10.1016/j.enconman.2020.113731