Short-Term Offshore Wind Speed Prediction Based on OWT-LSTM-ARIMA Combined Model

Abstract

The randomness and volatility of offshore wind speed make a single prediction model less accurate. In order to improve the prediction accuracy, this paper proposes a combined offshore wind speed prediction model based on the combination model of orthogonal wavelet transform (OWT) -Long -Short Term Memory Network (LSTM) -Autoregressive Integrated Moving Average (ARIMA) model. The OWT is first used to decompose the offshore wind speed series into low-frequency sub-series and high-frequency sub-series to reduce the impact of volatility. The LSTM model is used to predict the high frequency subseries, and the ARIMA model is used to predict the low frequency subseries. Then the prediction results are fused to form a complete prediction of the offshore wind speed series. In addition, K-Medoids algorithm is improved to simulate the extreme weather at sea represented by lightning to further prove the superiority of the combined model. The results show that the combined model has higher prediction accuracy and stronger stability than the single model.