A green hydrogen production model from solar powered water electrolyze based on deep chaotic Lévy gazelle optimization

Abstract

This paper presents a Deep Learning (DL) model designed for green hydrogen production using a solar-powered water electrolyzer. The model operates in four phases, beginning with the analysis of a solar radiation dataset and culminating in the prediction of green hydrogen production. A novel hybrid model, termed DeepGaz, is introduced to predict the solar energy required for hydrogen production. DeepGaz combines a new Chaotic-Lévy variant of the gazelle optimization algorithm (CGOA) with a recurrent neural network (RNN/LSTM) for hyperparameter optimization. To validate the performance of the proposed model, CGOA is first tested on the CEC2022 benchmark problems and compared with other advanced metaheuristic algorithms, with its accuracy further confirmed using Wilcoxon’s rank-sum statistical analysis. Subsequently, DeepGaz is applied to a solar-based green hydrogen dataset, optimizing key parameters such as solar radiation, temperature, wind direction, and speed, collected from the HI-SEAS weather station in Hawaii over a four-month period. The results show that DeepGaz significantly improves the prediction process, achieving an average daily hydrogen production of 15.5199 kg/day during the four-month study. The model exhibits strong potential in predicting green hydrogen production, excelling in computational time, convergence stability, and overall solution accuracy.

Keywords

• Green hydrogen
• Chaotic-Lévy gazelle optimization algorithm
• Artificial intelligence
• Deep learning
• Recurrent neural network
• Solar green energy