Particle swarm optimization algorithm with Gaussian exponential model to predict daily and monthly global solar radiation in Northeast China

Abstract

Reliable global solar radiation (R_s) information is crucial for the design and management of solar energy systems for agricultural and industrial production. However, R_s measurements are unavailable in many regions of the world, which impedes the development and application of solar energy. To accurately estimate R_s, particle swarm optimization (PSO) algorithm integrating Gaussian exponential model (GEM) was proposed for estimating daily and monthly global R_s in Northeast China. The PSO-GEM was compared with four other machine learning models and two empirical models to assess its applicability using daily meteorological data from 1997 to 2016 from four stations in Northeast China. The results showed that in different stations, the PSO-GEM with full climatic data as inputs showed the highest accuracy to estimate daily R_s with RMSE, RRMSE, MAE, R², and E_ns values of 1.045-1.719 MJ m^-2 d^-1, 7.6-12.7%, 0.801-1.283 MJ m^-2 d^-1, 0.953-0.981, and 0.946-0.977, respectively. The PSO-GEM showed the highest accuracy to estimate monthly R_s with RMSE, RRMSE, MAE, R², and E_ns values of 0.197-0.575 MJ m^-2 d^-1, 1.5-7.0%, 0.137-0.499 MJ m^-2 d^-1, 0.999-1, and 0.992-0.999, respectively. Overall, the PSO-GEM had the highest accuracy under different inputs and is recommended for modeling daily and monthly R_s in Northeast China.

Keywords: Empirical models; Gaussian exponential model; Global solar radiation; Machine learning models; Particle swarm optimization.