Machine learning-driven optimization of an integrated renewable energy system for power, water, and cooling with CO2 capture

Rising worldwide demands for sustainable freshwater, energy, and cooling are increasingly constrained by the exhaustion of fossil energies, surging greenhouse gas emissions, and intensifying water scarcity. Low-carbon systems can simultaneously provide clean energy and environmental benefits. This study presents a biomass-based hybrid design that produces electricity, cooling, desalinated water, and removes atmospheric CO2 in one structure. The model links biomass gasification, a SOFC (solid oxide fuel cell), a CO2 power unit, a cooling stage, a desalination-unit, and a carbon-capture process for sustainable performance. Its novelty lies in the tightly coupled design that addresses energy, environmental, and resource concerns concurrently, while leveraging machine learning–based surrogate models—Random Forest, Bayesian Gaussian Process Regression, Fine Gaussian Support Vector Machine, and Elastic Net Regression—linked with a hybrid Ant Lion Optimizer (ALO) algorithm for optimization. Thermodynamics have shown that the efficiency for energy is 76.3 % and exergy is 34.3 %. Results also show a capacity to produce net electricity of 113.2 kW, cooling of 214.72 kW, potable water at 23.79 × 106 kg/y, and CO2 at 1.47 × 106 kg/y, showing very good overall performance by the system. Exergoeconomics has confirmed that resources are being utilized efficiently economically and that there is room for improvement primarily in reducing exergy destruction in the major components. The hybrid ALO algorithm improves convergence stability and broadens solution diversity, surpassing conventional optimization methods. This study advances biomass-based energy conversion by establishing a scalable, cost-effective framework for integrated multigeneration, efficient resource use, and tangible gains in energy performance and environmental sustainability.