Analysis of ANN-based daily global horizontal irradiance prediction models with different meteorological parameters: a case study of mountainous region of India

Abstract

ABSTRACT Solar resource availability of a location depends on the local meteorological parameters. Therefore, the selection of suitable meteorological variable is highly desirable to develop the accurate solar irradiance prediction models. In present study, global horizontal irradiance (GHI) prediction models are developed using artificial neural network (ANN) with different combinations of meteorological parameters. A four-year station measured dataset consists of minimum temperature (), maximum temperature (), temperature difference (), GHI, extraterrestrial radiation (), and bright sunshine hours () have been employed to establish ANN models. Five types of ANN models (ANN-1 to ANN-5) are developed with 32 possible input combinations to determine the best input combinations to predict the daily GHI accurately. The achieved maximum correlation coefficient for ANN-1, ANN-2, ANN-3, ANN-4, and ANN-5 models are 0.9197, 0.9681, 0.9688, 0.9515, and 0.9457, respectively. The results revealed that ANN-2 and ANN-3 has shown best performance with the input combinations of [] and respectively. The proposed methodology is also used to assess the solar potential of the mountainous state of Uttarakhand, India, situated in foots of Himalayas, using the best ANN model. The obtained results suggest that Uttarakhand has good solar potential with annual GHI varies from 16.96 to 19.54 , which is sufficient to implement a broad range of solar applications in the region. The methodology proposed in this work can be utilized to develop solar irradiance prediction models for different locations where monitoring stations are not available.