High-resolution AI-based forecasting and techno-economic assessment of green hydrogen production from a hybrid PV/wind system at the regional scale
Salma Boujmiraz, Abdellatif El Ghazi, K. Bouziane, M. J. Sanjari, Omar Jellouli, Asmae Berrada
Abstract
This study presents a comprehensive framework that integrates high-resolution energy forecasting and techno-economic modeling to assess green hydrogen production potential in Flanders, Belgium. Using 15-min interval data from the Elia Group, four deep learning models (LSTM, BiLSTM, GRU, and CNN-LSTM) were developed to forecast regional photovoltaic (PV) and onshore wind energy generation. These forecasts informed the estimation of hydrogen yields and the evaluation of the levelized cost of hydrogen (LCOH) under different configurations. Results show that wind-powered hydrogen production achieves the lowest LCOH (6.63 €/kg), due to higher annual operating hours. Among electrolysis technologies, alkaline electrolysis (AEL) offers the lowest cost, while proton exchange membrane (PEMEL) provides greater flexibility for intermittent power sources. The hybrid PV-wind system demonstrated seasonal complementarity, increasing annual hydrogen yield and improving production stability. The proposed framework supports regional planning and highlights strategic investment opportunities for cost-effective green hydrogen deployment. • Integrated deep learning and techno-economics for green hydrogen assessment. • Linked forecasts with PEM electrolysis to estimate hydrogen output and costs. • Develops a scalable tool for regional green hydrogen planning and investment.