ABR处理硫酸盐有机废水的BP神经网络建模

通过厌氧折流板反应器(ABR)处理硫酸盐有机废水的实验数据对BP神经网络进行训练,建立了ABR处理硫酸盐有机废水的BPNN模型,通过测试对比,找出了较优训练函数为traingda,较优训练次数为1 900.利用分割连接权值法(PCW)对影响出水SO42-和COD的主要因素进行分析,结果显示进水COD、SO42-、pH、COD/SO42-和HRT对出水SO42-和COD均产生一定影响,其中进水pH对出水SO42-和COD的影响最大,相对重要性(RI)指数分别为30.79％和23.44％;并通过样本试验数据分别建立了对SO42-和COD去除率的限制因子仿真模型,为预测硫酸盐有机废水的厌氧处理过程提供指导.

Model and simulink of anaerobic baffled reactor treating sulfate organic wastewater based on back-propagation neural network

The back-propagation neural network(BPNN) trained with the data from the sulfate organic wastewater treatment of anaerobic baffled reactor(ABR) and a network model was built. The better training function and times were 'traingda' and 1 900, respectively. Partition connection weights (PCW) was adopted to analyze the dominant factors of effluent COD and SO₄^2-. The results showed that all of the factors (feed COD, SO₄^2-, pH, COD/SO₄^2- and HRT) had an influence on effluent COD and SO₄^2-. Nevertheless, the feed pH was the dominant factor, which relative importance (RI) were 30.79% and 23.44%, respectively. The model and simulink on restrictive factors for COD and SO₄^2- removal were built respectively, which can be used for prediction on sulfate organic wastewater treatment.