Prediction of Wastewater Plant Effluent COD Based on Dynamic Neural Network

Aiming at the problem of cost redundancy of biochemical tanks in municipal wastewater treatment plants, a method for accurately predicting the effluent chemical oxygen demand(COD) of biochemical tanks is proposed. Firstly, the data collected from the wastewater treatment plant are preprocessed. Secondly, a long short-term memory (LSTM) neural network model is established with influent water quality indices, aeration quantity and drug content as inputs and effluent COD as outputs. Thirdly, a dynamic variable weight algorithm is used to dynamically optimize the gating weights of the LSTM neural network. Finally, in order to verify the effectiveness of the method, the prediction results of the dynamic LSTM model are compared with the prediction results of the LSTM-based model. The results show that the sum of squares error of dynamic LSTM model is reduced by 16.15 percent compared with LSTM model. Therefore, the model established in this paper is effective and feasible for the prediction of wastewater indicators in biochemical tanks.