Techno-economic analysis and optimization of grid-connected hybrid renewable energy systems with hydrogen storage and machine learning-based solar forecasting

Industrial energy systems require reliable integration of renewable energy. Accurate solar irradiance forecasting enables optimal hybrid system design for sustainable operations. This study develops an integrated approach combining advanced direct normal irradiance (DNI) forecasting with optimal design of a grid-connected hybrid renewable energy system for industrial applications in Sanjiao, Zhongshan, China. Several machine learning models were evaluated for DNI prediction using hourly meteorological observations, with the Extended Long Short-Term Memory (xLSTM) model achieving superior performance with an R2 of 0.992. The model was trained on 2023 historical data to characterize solar resource patterns for system design optimization, distinct from operational weather forecasting, which is limited by the meteorological prediction horizon. The high-accuracy DNI predictions from the xLSTM model were subsequently integrated into the simulation framework to ensure reliable sizing and operational planning of the renewable energy components. The optimized energy configuration comprises 17,204 kW of solar photovoltaics, 12,260 kW of wind turbines, 89,989 kW h of lithium-ion batteries, and a hydrogen storage subsystem including a 2000 kW electrolyzer, a 3000 kg storage tank, and a 500-kW fuel cell. The system achieves a levelized cost of energy of $ 0.2837/kWh and a net present cost of $146,034,200 over a 20-year lifetime. Technical performance analysis demonstrates exceptional reliability, with an 87.2% renewable fraction and minimal unmet load and capacity shortages. The system generates 58,468,003 kW h annually from renewable sources, exports 3,533,928 kW h net energy to the grid, and maintains energy autonomy with only 6.22% grid dependency. The dual-storage strategy effectively manages renewable intermittency, with batteries and hydrogen storage ensuring long-duration energy security. This comprehensive framework demonstrates the technical and economic viability of transitioning industrial energy systems toward high renewable penetration while maintaining grid stability and operational reliability.