Research on Forecasting Method for Effluent Ammonia Nitrogen Concentration Based on GRA-TCN

Abstract

Aiming at the characteristics of high coupling degree, strong nonlinearity, and serious time delay in the measurement of ammonia nitrogen concentration in wastewater treatment process (WWTP), a prediction model of ammonia nitrogen concentration based on gray relational analysis (GRA) and time convolution network (TCN) was proposed. Firstly, based on the relevant water quality parameters collected in WWTP, the grey correlation analysis method was used to find out other characteristic variables highly related to the ammonia nitrogen concentration. Then, a new group of multivariate time series data was constructed by using the sliding window method. Finally, based on the advantages of the time convolution network in processing time series data, such as simple, flexible, and easy to parallel, the constructed time series data were modeled to predict the concentration of effluent ammonia-nitrogen. To verify the validity of the model, the predicted results were compared with the other four models. The experimental results show that the ammonia-nitrogen concentration prediction model based on GRA and TCN has good prediction performance, which is helpful to realize the accurate prediction of effluent ammonia-nitrogen concentration. At the same time, it can also provide timely and effective guidance for the control and optimization of the wastewater treatment process.