光催化降解对硝基苯胺的模拟研究

A simulating study of photocatalytic degradation of p-Nitroaniline

以四因素五水平中心组合实验的27组实验数据为训练集，按U12（12×4^3）设计的12组实验数据为预测验证集，采用BP神经网络（BPNN）和二次回归模型进行建模比较，研究[TiO2]、Co、pH值和t4个变量对p-NA光催化降解率DC的影响。结果表明BPNN模型优于二次回归模型，该模型对预测验证集的预测结果为相关系数R=0.9474，平均相对误差绝对值MRE（％）=15.2。由BPNN模型的权值计算出[TiO2]、Co、pH值和t4个变量对p-NA的DC的贡献大小分别为31．96％，28．47％，9．2％，30．37％，用建立的BPNN模型模拟分析[TiO2]、pH值和Co3个变量对DC的变化趋势的影响，根据模拟分析得出该体系的优化实验条件为C0=20mg／L、[TiO2]=2.35g／L、pH=5.0，降解35min的DC为97．56％，实验验证结果DC为95.98％，实验值与模拟值相对误差仅为-1．60％。

Taking the central composite experimental design of four factors and five levels of 27 data as the training set and the experimental data of U12（12×4^3） as the predicted validation set, BPNN model and quadratic regression model are compared in order to discuss the influence of four variables （the quantity of TiO2, p- nitroaniline（ Co ）, pH and t） on the degradation rate（ DC% ）. The results show that BPNN model is better than quadratic regression model. The results of BPNN show that the predicted correlation coefficient R is 0. 9474 and the mean relative error between the predictive value and experimental value is 15.2%. The contributions of four variables of [ TiO2 ] , Co, pH and t to DC% are computed based on the connection weights of BPNN which are 31.96% , 28. 47% , 9. 2% , 30. 37% respectively. The influences of three variables of [ TiO2 ] , Co and pH on DC% are discussed. The optimized experimental condition of the system is obtained ： CO = 20 mg/L, [ TiO2 ] = 2. 35 g/L and pH = 5.0. Under the optimized experimental condition in 35-minute degradation, the experimental value of DC% is 95.98% and the predictive value is 97.56%. The relative error is - 1.60%.