A novel hybrid model based on artificial neural networks for solar radiation prediction

In this paper, a new hybrid model combining the SOM-OPELM with time series strategies is presented for predicting the global solar radiation on the horizon. In this model, the SOM divides the original data into distinct clusters and the OPELM establishes the prediction model. Subsequently, three population time series strategies, (i.e. Recursive strategy, DirRec strategy and MISMO strategy) are adopted to accomplish the multi-step prediction. A comparison between the proposed SOM-OPELM model and other conventional methods is carried out to demonstrate its efficiency and estimation performance. The simulation results show that the proposed SOM-OPELM model with DirRec strategy or MISMO strategy outperforms the following models: Recursive-BP, DirRec-BP, MISMO-BP, Recursive-SOM-OPELM and ARIMA.