The invention discloses a
water index prediction method based on
sewage treatment, which comprises the following steps of: stirring and standing water, dividing regions and depths, establishing a local position
water supply amount model, adjusting water inlet indexes for the first time, and establishing a logarithmic
function model to predict the water inlet indexes of different depths. Finally, five models are established to predict the PH value, the COD, the
ammonia nitrogen content, the
total phosphorus content and the
total nitrogen content respectively, and predicted values are displayed after time correspondence. According to the method, firstly,
water supply and layered modeling are performed after standing, then dynamic changes of
water quality indexes are accurately reflected in combination with the vertical distribution characteristics of fluid after standing, and a logarithmic function is constructed based on the law that the concentration is attenuated along with time during
precipitation and
diffusion of pollutants to predict water inlet indexes in the
depth direction to reduce the sampling frequency. By means of grid dynamic correction and weighting of the displacement proportion by means of the arithmetic mean value, sampling deviation caused by flow disturbance is reduced, prediction robustness is improved, it can be guaranteed that the quality of
sewage is relatively uniform through the scheme, then the optimized water inlet index is obtained by weighting the displacement proportion by means of the mean value of the water inlet indexes of all the areas, and the prediction accuracy is improved. And calculating the optimized water inlet indexes through the constructed logarithmic function to obtain water inlet indexes of different depths, and inputting the water inlet indexes of different depths into the set five groups of models for calculation and prediction to obtain water outlet indexes.