基于BP神经网络的多壁碳纳米管复合CdS-TiO_2光催化剂优化合成研究

Optimizing the preparation of CdS-TiO_2/MWCNTs photocatalyst based on BP neural network

基于CdS-TiO2/多壁碳纳米管(MWCNTs)光催化剂降解甲苯的正交实验数据,采用反向传播(BP)神经网络训练并建立了光催化剂合成条件设计的神经网络模型。以正交实验确定的4个主要影响因素作为输入层参数,以甲苯降解率作为输出层参数,将全部实验数据分为训练样本集和预测样本集。运行网络,系统误差为0.000 724,网络预测值与实验数据值相关系数达到0.989,说明该网络具有较好的训练精度及泛化能力。并利用训练好的神经网络预测得到CdS-TiO2/MWCNTs光催化剂的最佳合成条件:焙烧温度为460℃,MWCNTs复合量为1.5%(质量分数),活性组元摩尔比(TiO2/CdS)=80∶1,水加入量为12%(体积分数)。

The orthogonal experiment data of toluene photodegradation by CdS-TiO2/multi-wall carbon nano- tube （CdS-TiO2/MWCNTs） were used to train the Back Propagation （BP） neural network,and the established net- work model was applied to predict the optimal preparation conditions of CdS-TiO2/MWCNTs photocatalyst. The key influential factors on activities of photocatalyst were regarded as characteristic input vectors, and toluene removal as output vectors. The experimental data were divided into train group and prediction group. Running the BP neural net- work,the system error was 0. 000 724, and the correlation degree between network predictive data and experimental data was 0. 989,showing its perfect precision and the generalization ability. Predicted by well-trained BP neural net- work,the optimal preparation conditions of the CdS-TiO2/MWCNTs photocatalyst were obtained as follows：the calci- nation temperature was 460℃ ,the mass ration of MWCNTs was 1.5% ,the molar ration of TiO2 to CdS was 80 ： 1, and the volume ratio of water was 12%.