Mixed Model Of Phosphorus Removal In Compound Constructed Wetland

With the rapid economic development and population growth, urban sewage continue to produce, artificial wetlands as an environmentally-friendly sewage treatment technology has been widely used. But wetland still design and prediction based on black box, it can’t be accurate and reliable evaluate the operating results of wetlands and also limit the development of wetland technology. In order to promote the artificial wetland technology it needs a construction of a simple,accurate and reliable forecasting model, therefore carried out the following:1. Analysis of the results of previous research, understand wetland structure, characteristics, pollutant removal mechanism and influencing factors, so we select TN (total nitrogen), TOC (total organic carbon) which has few reportsas the study factors.2. Treat FengHuang River wetland(in Cheng Du) as the research object, analysis the effect of TN and TOC to phosphorus removal and subsystems’ contribution by measuring the actual operating results. Found that the first biological pond has the highest removal rate, following is the second biological pond, the plant sand filter, second plant-gravel bed is the lowest, while the first plant-gravel bed increase phosphorus content of wastewater; the whole removal amount of TP is in a stable state; effluent concentration of phosphorus increases with the influent concentration of phosphorus; the content of TN, TOC reduced with content of total phosphorus. According to the correlation coefficient and the coefficient of determination, legend effluent phosphorus concentration and influent concentration of phosphorus, TOC and TP, TN and TP has a strong linear correlation.3. Using Monod model to simulate the biochemical reactions of wetland, combining radial basis function (RBF) artificial neural network model to consider the impact of TN, TOC, combining the advantages of both, work together to build a Monod-RBF mixed model of constructed wetland. By comparison with separate RBF neural network model and mechanism model, the absolute error of mixed model<mechanistic model<RBF model; the relative error of mixed model<mechanistic model<RBF model;the standard error of mixed model (SEP)<mechanistic model< RBF model, indicating mixed model has higher prediction accuracy,it can predict the treatment effect of constructed wetland better.