BP神经网络在膜分离处理废水中的应用

为能更好地预测工艺条件对膜分离过程的影响,运用BP神经网络技术建立输入变量为压差、流速、浓度和温度,输出变量为膜通量的预测模型。通过大量实验数据训练预测模型,得到的网络模型整体误差平方和仅为0.014 5;计算值与模拟值相比,10组不同条件的膜通量平均预测误差仅为1.1,证实了所建立的BP神经网络膜通量预测模型与实验值吻合程度较好,有较好地预测能力。在此基础上进一步考察了工艺参数对膜分离过程的影响。

An Application to BP Neural Network and Membrane Separation Technology in Wastewater Treatment

In waste water treatment process,prediction model was built by BP neural network in order to predict the influences of technological conditions on membrane separation process.The input variables were transmembrane pressure,cross-flow velocity,concentration and temperature with output variable membrane flux.The prediction model was trained by mass experimental data,the overall square sum of error was only 0.014516;compared with calculated values,average error of membrane flux under 10 different conditions was only 1.1.The results showed that membrane flux prediction model on BP neural network indicated high accuracy with experimental data.On the basis of the above studies,the influences of technological parameters on membrane separation process were also studied.