Short‐Term Ensemble Insolation Forecasting Method Using Parallelized NE

ABSTRACT In this paper, the authors propose an ensemble forecasting method using multiple individuals for short‐time‐ahead (1 h ahead) insolation forecasting by using neuroevolution (NE), in which a genetic algorithm is applied to the learning algorithm of a neural network for insolation. Although the method improves the accuracy compared to a single forecast, NE has a problem that the training time is long. In order to solve this problem, the authors propose a parallelization method of GPU processing for short‐time‐ahead forecasting and try to solve the problem by parallelizing the GPU.