Enhanced Prediction of Solar Irradiance Using a Hybrid Approach Based on the Crow Search Algorithm and Extreme Learning Machine Network

Solar energy has a higher degree of volatility due to climatic constraints and scenarios. The efficient and successful deployment of solar energy necessitates an accurate and robust prediction method for predicting solar irradiance (GHI: Global Horizontal Irradiance). Inappropriate selection of Extreme Learning Machine network (ELMN) parameters creates the generalization issue, computational burden and unnecessary complexity. To address the issue of optimizing ELMN parameters. This research work addresses the issue with the development of a hybrid prediction approach (CSA-ELMN) combination of the Crow Search Algorithm (CSA) and Extreme Learning Machine Network (ELMN). The novel aspect of this investigation is using a crow search algorithm during the extreme learning machine training phase to optimize synaptic connection weights, bias and hidden layer neurons, which have been successfully evaluated in the Solar Irradiance (GHI) predictions application. Four statistical indices, including the mean square error (MSE), mean absolute percentage error (MAPE), root mean square error (RMSE), and mean relative error (MRE), were computed to assess the proposed hybrid prediction model (CSA-ELMN). The findings of the CSA-ELM approach shows that it improves GHI prediction precision compared to other traditional and hybrid approaches.