Integrating Machine Learning for Predictive Modeling of Carbon Sequestration in Mixed Microalgal Systems

Microalgal carbon sequestration offers an energy-efficient solution to global carbon emissions. Mixed microalgae outperform pure cultures in adaptability and carbon fixation. This study carried out mixed microalgae cultivation experiments under different experimental conditions, focusing on examining the effects of five key factors: inoculation concentration, initial inoculation ratio, nitrate concentration, CO₂ gas flow rate, and CO₂ concentration. In the single-variable experiment where the CO₂ gas flow rate was adjusted, the maximum carbon sequestration of 5.35 g/L was achieved. An experimental data set of carbon sequestration performance characteristics of mixed microalgae under five culture conditions. The carbon sequestration performance of mixed microalgae was predicted using 13 machine learning models. The results showed that the linear model (R ² = 0.782) performed the poorest, followed by the nonlinear model (R ² = 0.941). The ensemble model (R ² = 0.950) significantly outperformed both, with random forest exhibiting the best performance.