基于递归神经网络的焦化废水水质预报

采用铁碳电池预处理、厌氧－好氧－好氧－缺氧多级SBR工艺处理焦化废水实现了稳定的亚硝化反硝化生物脱氮。好氧，缺氧反应器的水质波动大，若能对其水质进行准确预报，对于指导生产操作及工艺流程实现计算控制具有重要意义。本文建立了一个三层递归神经网络模型，对一级好氧、二级好氧、缺氧反应器的主要水质指标实现了准确预测，预报平均相对误差分别为2．86％，4．99％，4．2％。

The Prediction of Coke-plant Wastewater Quality Based on Recurrent Neural Network

Stable and good results are achieved through the treatment of coke-plant wastewater by pretreated iron scrap-anaerobic -oxic-oxic-anoxic SBR process with shortened nitrification-denitrification pathway. The quality of water is unstable in oxic and anoxic reactors. If the quality of wastewater can be predicted, operation will become much easier and that the computer controlled process will be easily realized. A three-layer simple recurrent neural network model is proposed to accurately predict the quality of wastewater in the oxic and anoxic reactors. The individual fractional errors are 2.86%, 4.99% and 4.2%.