A study on the UKFNN-based online detection of effluent COD in water sewage treatment

Aiming at the nonlinear sewage system, this paper is prepared using the unscented Kalman filter neural network (UKFNN) for online detection of the effluent chemical oxygen demand (COD) concentration in wastewater treatment. Six environmental factors (e.g. dissolved oxygen, ammonia nitrogen value, PH value) that may pose effect on the real-time monitoring of COD concentration are considered to get the information on the COD concentration changing with various environmental factors. By comparing the BPNN model, experimental results showed that the soft measurement model of UKFNN could be faster and more accurate in prediction, where the correlation coefficient of the predicted value and the actual value was 0.991, with the maximum relative error of only 4.7%, being able to achieve on-line detection of COD.