基于BP神经网络的Cu-Ce掺杂TiO2光催化剂性能预测组合模型研究

采用环境测试舱模拟可见光下的室内环境,以甲醛气体的光催化降解为探针反应,评价了Cu-Ce/TiO2光催化剂的光催化活性及对甲醛气体的去除效果。利用指数平滑-神经网络ES-BP组合模型对Cu-Ce/TiO2光催化剂的性能做预测分析。结果表明:经过Cu-Ce/TiO2光催化剂处理后细木工板中甲醛释放浓度明显降低,平均光催化降解甲醛气体效率为42.8%;ES-BP组合预测模型在Cu-Ce/TiO2光催化剂的性能预测中取得了较好的效果,平均绝对误差为-0.00011mg/m3,平均相对误差为-0.317%;ES-BP组合预测模型实现了BP神经网络模型和指数平滑模型的优势互补,提高了对数据长期预测的准确性。

Research on Hybrid Prediction Methods for Cu-Ce-doped TiO2 Photocatalytic Performance Based on BP Neural Network

The photocatalytic activities of Cu-Ce/TiO₂ photocatalyst were determined and removal effect of formaldehyde was investigated by the photocatalytzed decomposition of gas formaldehyde in the environmental chamber using a visible lamp. The BP-ES hybrid model based on BP neural network and exponential smoothing was proposed to predict the performance of Cu-Ce/TiO₂ photocatalyst. The results showed that after Cu-Ce/TiO₂ photocatalyst treatment, formaldehyde emission concentration from block board was decreased obviously, and the average efficiency for photocatalytic degradation of formaldehyde gas was 42.8%. BP-ES hybrid model achieved good results in the perfor-mance prediction of Cu-Ce/TiO₂ photocatalyst,as the average absolute error was -0.00011 mg/m³ and the average relative error was -0.317%. BP-ES hybrid model can realize the complementary advantages of the BP neural network model and exponential smoothing model, and improve the accuracy of data long-term forecast.