摘要
考察了0．25～0．3mm规格的粉末活性炭在不同活性炭投加量、溶液浓度、时间和温度的条件下对亚甲基蓝溶液的静态吸附情况。利用MATLAB自行设计了BP神经网络预测系统，分别建立了活性炭投加量、溶液浓度、时间和温度与亚甲基蓝去除率的复杂的非线性关系。用实验数据对该神经网络模型进行了训练，经518次达到0．0001的精度要求，并对测试数据进行了预测。结果表明，预测值与实测值的绝对误差仅为-0．0042和0．0014，进而说明该BP神经网络模型在本研究系统的建立是成功的。
The static solution of the 0.25~0.3mm specifications powder activated carbon on methylene blue adsorption under different conditions,such as carbon dosage,concentration,time and temperature factors is investigated.The complex nonlinear relationship of the dosage of activated carbon,solution concentration,time and temperature with methylene blue removal is established by BP neural network prediction system based on MATLAB.The neural network model is trained by using the experimental data,and the accuracy meets 0.0001 after trained for 518 times and the test data is proved right.The results indicate that the predicted and measured values are very close,and the error is only-0.0042 and 0.0014.Respectively,the BP neural network model system established in this study is successful.
出处
《材料导报》
EI
CAS
CSCD
北大核心
2010年第24期73-75,共3页
Materials Reports
基金
广东省工业攻关项目基金资助(2005B10301051)
