讨论基于BP神经网络一体化AmOn污水处理工艺数学模型的构建。在分析神经网络和一体化AmOn污水处理工艺各自特点的基础上,建立了具有11个输入层节点、5个输出层节点以及包含两层隐含层的BP神经网络,并确定了各隐含层的节点数、网络的权值和阈值。对模型的验证结果表明:BP神经网络模型基本反映了AmOn一体化工艺的主要特点,网络预测出水结果与实测结果相差不大,神经网络模型具有较好的泛化能力。 Discuss the construction of mathematical model of integrated AmOn wastewater treatment based on BP neural network. Based on the analysis of the respective characteristics of the neural network and integrated AmOn wastewater treatment process, a BP neural network with 11 input nodes, 5 output nodes and two hidden layers is established and each hidden layer is determined Of nodes, network weights and thresholds. The results of the model verification show that the BP neural network model basically reflects the main features of the AmOn integrated process. The results of the network prediction water and the measured results are similar, and the neural network model has a good generalization ability.