Prediction Model of Wastewater Pollutant Indicators Based on Combined Normalized Codec

Effective prediction of wastewater treatment is beneficial for precise control processes. The nonlinearity pollutant indicators such as chemical oxygen demand (COD) and total phosphorus (TP) makes the model difficult to fit has low accuracy. classical deep learning methods have been shown perform nonlinear modeling. However, there are enormous numerical differences between multi-dimensional data in problem treatment, COD above 3000 mg/L TP around 30 mg/L. It will make current normalization challenging handle effectively, leading training failing converge gradient disappearing or exploding. This paper proposes a multi-factor based on learning. consists combined layer codec. combines advantages three calculation methods: z-score, Interval, Max, which can realize adaptive processing data, fully retain characteristics finally cooperate with codec learn output results. Experiments show that proposed overcome complex predicting industrial achieve better accuracy than models.