Comparative Study of Short-Term Wind Speed Forecasting Techniques Using Artificial Neural Networks

摘要

This paper focuses on the importance of wind forecasting and comparison of two different forecasting schemes using artificial neural network approach. Types of forecasting include feed-forward network models using standard back propagation technique and recurrent neural network models with inherent memory for any given data. In this study, how local memory and relevant inputs make recurrent neural networks more suitable for time-series prediction than normal feed-forward networks is shown. And also for accurate forecasting and better energy trading, fine tuning of present techniques is required. Therefore, LSTM models are implemented which are a part of recurrent neural networks. Finally, the results are measured in terms of mean-squared error, an error function which calculates the difference between actual and model outputs. It was found that LSTM models were more suitable for short as well as long term time-series forecasting as compared to RNN model.