Intelligent models as novel tools for optimizing ultrasonication-ozonation technique in PAH-contaminated soil remediation
摘要
Polycyclic aromatic hydrocarbons (PAHs) pose serious risks to soil and human health, necessitating effective remediation strategies. This study investigates the efficiency of hybrid ultrasonication-ozonation treatment for the removal of Anthracene (ANT) and Phenanthrene (PHE) from contaminated soils, outperforming individual methods. Experimental results demonstrated that the hybrid method achieved superior removal efficiencies, with maximum values of 70.4% for ANT and 92.0% for PHE at an initial concentration of 350 mg/kg. In contrast, pure ultrasonication reached lower removal rates of 38.9% for ANT and 45.8% for PHE at 390 mg/kg, while ozonation alone achieved 68.4% and 77.1%, respectively. These results highlight the synergistic effect of ultrasonication and ozonation in enhancing PAHs degradation. Machine learning algorithms were developed to predict PAHs removal efficiencies, with the Boosting model achieving the highest predictive accuracy (R² = 0.999 in training and 0.865 in testing datasets). SHAP analysis revealed that Ozone concentration, ultrasonic power, water volume, and initial contaminant concentration were the most influential features. Non-linear relationships and thresholds were observed, particularly for water volume and ultrasonic power, indicating optimal ranges beyond which removal efficiency declined. The study also underscores the critical role of soil texture, especially silt content, in modulating remediation outcomes. These findings provide valuable insights for optimizing PAH removal through advanced oxidation processes.