【摘要】 抗生素是环境中广泛存在的药物,其在水环境的迁移和分布主要受到沉积物吸附行为的影响.本文首先考察了高岭土、黑碳、腐殖酸、pH、Ca²⁺等5种影响因素对两种不同抗生素在沉积物上吸附强度的单因素影响,然后应用中心复合实验设计考察了其复合影响,并利用实验所得数据,分别拟合和验证了基于线性方程和BP神经网络的两种抗生素吸附模型,通过对比拟合和验证结果分别得到适用于两种抗生素的吸附模型.实验结果表明,pH和Ca²⁺对两种抗生素在沉积物上的吸附容量影响显著,而高岭土、黑碳和腐殖酸则影响较小.模型模拟结果表明,神经网络模型拟合程度和精度均优于线性方程模型;且交叉验证结果表明,利用不同组数据进行训练,神经网络模型拟合均取得了优于线性方程的拟合结果.因此,在所考察的因素和浓度范围内,神经网络模型较好地预测了沉积物中抗生素的吸附行为.更多还原

【Abstract】 Antibiotics are widely presented in the environment. Their migration and distribution in the aquatic environment are mainly affected by the adsorption behavior of sediments. In this paper, the effects of kaolin, black carbon, humic acid, pH and Ca²⁺ on the adsorption strength of two different antibiotics on sediments were investigated. Then the central composite design was used to investigate the composite effect. According to the experimental data, multi linear equations and BP neural network were used to fit and validate two antibiotic adsorption models. The adsorption models for two antibiotics were obtained by comparison fitting and verification results. The results showed that the effect of pH and Ca²⁺ on the adsorption capacity of the two antibiotics on the sediment were significant, while the effect of kaolin, black carbon and humic acid on the adsorption capacity of the two antibiotics on the sediment were less. The model simulation results showed that the degree and accuracy of the neural network model were better than the multi linear equation, and the cross validation results also indicate that the neural network model stimulate was better than the multi linear equation. Therefore, the neural network model was able to predict the adsorption behavior of antibiotics in sediments within the factors and concentrations examined reasonably.更多还原