Global solar radiation prediction over North Dakota using air temperature: Development of novel hybrid intelligence model

Accurate solar radiation (SR) prediction is one of the essential prerequisites of harvesting solar energy. The current study proposed a novel intelligence model through hybridization of Adaptive Neuro-Fuzzy Inference System (ANFIS) with two metaheuristic optimization algorithms, Salp Swarm Algorithm (SSA) and Grasshopper Optimization Algorithm (GOA) (ANFIS-muSG) for global SR prediction at different locations of North Dakota, USA. The performance of the proposed ANFIS-muSG model was compared with classical ANFIS, ANFIS-GOA, ANFIS-SSA, ANFIS-Grey Wolf Optimizer (ANFIS-GWO), ANFIS-Particle Swarm Optimization (ANFIS-PSO), ANFIS-Genetic Algorithm (ANFIS-GA) and ANFIS-Dragonfly Algorithm (ANFIS-DA). Consistent maximum, mean and minimum air temperature data for nine years (2010-2018) were used to build the models. ANFIS-muSG showed 25.7%-54.8% higher performance accuracy in terms of root mean square error compared to other models at different locations of the study areas. The model developed in this study can be employed for SR prediction from temperature only. The results indicate the potential of hybridization of ANFIS with the metaheuristic optimization algorithms for improvement of prediction accuracy. (C) 2020 The Authors. Published by Elsevier Ltd.

Automated summary

The study introduced a novel intelligence model by hybridizing ANFIS with two metaheuristic optimization algorithms for accurate global solar radiation prediction. The proposed model outperformed other models by 25.7%-54.8% in terms of accuracy, showing potential for improvement in prediction accuracy through hybridization.