Wind Speed Forecasting using Hybrid Model of CNN and LSTM with Wavelets
Kaung Myat San, Jai Govind Singh, K. Natarajan
Abstract
Accurate prediction of wind speed is essential for wind power generation contributing to the stability and reliability of the power system. This paper proposes a one-step-ahead wind speed prediction using a wavelet decomposition-based hybrid deep learning model. The hybrid model is developed using convolutional neural networks (CNN) and long short-term memory (LSTM) networks. The historical wind speed, temperature, and relative humidity data of Mandalay and Meiktila, Myanmar, is filtered using wavelets before being applied with deep learned networks. The filtered features are then used to forecast the wind speed using the CNN-LSTM model. The performance of the proposed hybrid model is compared with benchmark models, namely CNN and LSTM, with and without wavelets. The empirical study shows that the proposed hybrid model CNN-LSTM with wavelet decomposition outperforms other benchmark models.