近年來全球風力發電發展逐年攀升，世界風能協會指出2020年全球陸域及離岸風電，總裝置容量超過743GW。目前我國政府也積極投入在風能上的發展，並在新能源政策上規劃風力發電，目標為2025年達成6.9 GW，其中陸域風電1.2 GW，離岸風電約5.7 GW。台灣在地理環境上擁有良好的風場，位於桃園至雲林沿海一帶常有東北及西南季風吹襲，但是由於風的不穩定性及間歇性，常導致發電效率不佳，因此，風速的準確預測對於電網的穩定運行具有相當重要意義。
本研究採用了一種將經驗模態分解（Empirical Mode Decomposition, EMD）、k-means聚類算法與長短記憶神經網路（Long Short-Term Memory, LSTM）相結合的風速預測組合模型來解決該問題。該模型先執行k-means作為數據分群方法，將數據拆分為具有相似的組群;在使用EMD方法對風速序列進行分解為一定數量不同頻率的子序列，以降低序列的複雜性和非平穩性，而 LSTM模型用於預測各不同頻率的子序列，最後將各個子序列的預測結果結合起來，得到最終的預測結果。
為了評估所提出模型的擬合能力，本文比較依季節分類方式，循環類神經網路(RNN)及長短記憶神經網路(LSTM)等神經網路模型，實驗結果證明在使用測試數據中所採用的k-EMD-LSTM混合模型比單一模型MAPE提高了約45%，從接近11%下降到大約5%左右，表明使用k-means分類與EMD方法可以有效提高預測精度。

In this paper adopt a combined wind speed prediction model that combines Empirical Mode Decomposition (EMD), k-means clustering algorithm and Long Short-Term Memory (LSTM) neural network. The model first implements k-means as a data grouping method to split the data into groups with similar characteristics after using the EMD method to decompose the wind speed series into a certain number of subsequences with different frequencies to reduce the complexity of the series and non-stationary The LSTM model is used to predict subsequences of different frequencies, and finally the prediction results of each subsequence are combined to obtain the final prediction result.
In order to evaluate the fitting ability of the proposed model, this paper compares neural network models such as seasonal classification, Recurrent Neural Network (RNN) and Long Short-Term Memory neural network (LSTM). Get the best performance.

[1]工業技術研究院，風力發電單一服務窗口，潛力場址，https://www.twtpo.org.tw/offshore_show.aspx?id=963。
[2]全國法規資料庫，離岸風力發電示範獎勵辦法，101年7月3日，https://law.moj.gov.tw/LawClass/LawAll.aspx?pcode=J0130063。
[3]經濟部能源局，能源轉型白皮書，109年11月18日。
[4]林廷融，2021年世界風能報告風能在實現淨零排放的道路上扮演之角色，工業技術研究院綠能與環境研究所，2021年9月。
[5]M. S, Roulston, D. T. Kaplan, J. Hardenberg, and L. A. “Smith, Using Medium-Range Weather Forcasts to Improve the Value of Wind Energy Production,”Renewable Energy, 28(2003), 585–602.
[6]X. Y. Wu., B. Q. Wu, and Q. J. Q. Zhu, “Data-Driven Wind Speed Forecasting Using Deep Feature Extraction and LSTM,”IET Renewable Power Generation, 13(12), 2062–2069. 2019
[7]P. Chen, T. Pedersen, B. Bak-jensen, and Z. Chen,“ARIMA-Based Time Series Model of Stochastic Wind Power Generation,”IEEE Transactions on Power Systems, 25(2), 667 – 676. 2010
[8]H. Rodriguez, M. Medrano, L. M. Rosales, G. P. Pe˜nu˜nuri, and J. J. Flores,“Multi-Step Forecasting Strategies for Wind Speed Time Series,” 2020 IEEE International Autumn Meeting on Power, Electronics and Computing (ROPEC 2020). Ixtapa, Mexico.
[9]S. Liang, L. Nguyen, and F. Jin,“A Multi-Variable Stacked Long-Short Term Memory Network for Wind Speed Forecasting,”2018 IEEE International Conference on Big Data (Big Data). Seattle, WA, USA
[10]Y. Fu, W. Hu, M. Tang, R. Yu, and B. Liu,“Multi-Step Ahead Wind Power Forecasting Based on Recurrent Neural Networks,”2018 IEEE PES Asia-Pacific Power and Energy Engineering Conference (APPEEC) (pp. 217-222). Kota Kinabalu, Malaysia. Oct .2018
[11]Y. Zhou, F. Chang, F. L. Chang, I. Kao, and Y. Wang,“Explore a Deep Learning Multi-Output Neural Network for Regional Multi-Step-Ahead Air Quality Forecasts,” Journal of Cleaner Production, 209(2019), 134–145.
[12]H. Wang, Y. Zhang, J. Mao, and H. Wan,“A Probabilistic Approach for Short-Term Prediction of Wind Gust Speed Using Ensemble Learning,” Journal of Wind Engineering & Industrial Aerodynamics, 202(2020), 104198. 2020
[13]X. Liu, H. Zhang, X. Kong, and K. y. Lee,“Wind Speed Forecasting Using Deep Neural Network with Feature Selection,”Neurocomputing, 397(2020), 393–403. 2020
[14]A. K. Dubey, A. Kumar, V. García-díaz, A. K. Sharma, and K. Kanhaiya,“Study and Analysis of SARIMA and LSTM in Forecasting Time Series Data,”Sustainable Energy Technologies and Assessments, 47(2021), 101474. 2021
[15]G. Chen, B. Tang, X. Zeng, P. Zhou, P. Kang, and H. Long ,“Short-Term Wind Speed Forecasting Based on Long Short-Term Memory and Improved BP Neural Network,”International Journal of Electrical Power and Energy Systems, 134(2022), 107365. 2022
[16]Y. Dong, X. ma, C. ma, and J. wang,“Research and Application of a Hybrid Forecasting Model Based on Data Decomposition for Electrical Load Forecasting,” Energies, 9(12), 1050. 2016
[17]M. Liu, L. Ding, and Y. Bai,“Application of Hybrid Model Based on Empirical Mode Decomposition, Novel Recurrent Neural Networks and the ARIMA to Wind Speed Prediction,”Energy Conversion and Management, 233(2021), 113917. 2021
[18]Z. Tian, and H. Chen,“Multi-Step Short-Term Wind Speed Prediction Based on Integrated Multi-Model Fusion,”Applied Energy, 298(2021), 117248. 2021
[19]X. Chen, J. Zhao, X. Jia, and Z. Li,“Multi-Step Wind Speed Forecast Based on Sample Clustering and an Optimized Hybrid System,” Renewable Energy, 165(2021), 595–611.
[20]中央氣象局105年氣候年報, 2017年09月22日,[Online]. Available: https://www.cwb.gov.tw/V8/C/A/yearpaper.html.
[21]中央氣象局106年氣候年報, 2018年10月11日,[Online]. Available: https://www.cwb.gov.tw/V8/C/A/yearpaper.html
[22]中央氣象局107年氣候年報, 2019年11月04日,[Online]. Available: https://www.cwb.gov.tw/V8/C/A/yearpaper.html
[23]J. Yan, K. Li, E. Bai, J. Deng, and A. M. Foley,“ Hybrid Probabilistic Wind Power Forecasting Using Temporally Local Gaussian Process,”IEEE Transaction on Sustainable Enegy, 7(1), 87–95. 2016
[24]H. Liu, H. qi Tian, and Y. fei Li,“Comparison of New Hybrid FEEMD-MLP, FEEMD-ANFIS, Wavelet Packet-MLP and Wavelet Packet-ANFIS for Wind Speed Predictions,”Energy Conversion and Management, 89, 1–11. 2015
[25]E. Huang Norden, Shen Zheng, et al,“The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis,”Proceedings of the Royal Society 454 (1971) (1998)