Wind/solar power forecasting using improved LSTM neural networks

Abstract

Nowadays, new energy become more and more important not only for industry but also for our citizens. How to forecast the wind and solar power correctly is also necessary for power plant. In this dissertation, four kinds of forecasting system based respectively on NARX model, BP neural network model, RNN model, and LSTM neural network model are described and the performance of these model are compared. It is shown that the forecast result of LSTM model is much better than NARX model and other models. With a very small MSE, the LSTM model is really suitable for wind power and solar power forecasting.