Short-term Global Horizontal Irradiance Prediction Based on Deep Echo State Network

Abstract

The prediction of global horizontal irradiance have a great impact on the stability and economic benefits of photovoltaic (PV) power generation. In this paper, we adopt the method of Deep Echo State Network (DESN) to predict the global horizontal irradiance in different areas one hour in advance. Under the same conditions, the results show that DESN are better than BP, SVM, ESN methods in the prediction accuracy. Experiments show that the proposed models show superior ability in predicting solar irradiance and have great application potential in power grid integration.