Research Of Pollutant Removal In Tidal Flow Constructed Wetlands Based On Different Inflows Way

At present, water pollution is more and more serious and natural wetland purification ability is limited. Constructed wetland has many advantages, such as efficient processing capacity and low cost, so it gradually become the hot spot research. But it has low oxygen capacity and affected by many factors, so seek to strengthen technical method of removal efficiency of constructed wetland is required.As a new type of constructed wetland ecosystem, tidal flow constructed wetland(TF-CW) has been widely cited in for the research field of pollutant removal in recent years. In this paper, to study the pollutant removal efficiency and main removal factors in TF-CW, we established four TF-CW simulators(continuous flow wetland as a control(A); compared idle/response times 1:1(B), 1:2(C) and 2:1(D) for respectively). We studied pollutant removal effect and the change with depth, and use redundancy analysis screening the main influencing factors. All main factors would be imported artificial neural network model for raining and validation so as to forecast the effluent concentration of pollution. Main reaults obtained as follow:1. Average TN removal rate was 82.41±4.84% for A, and 84.82±5.09%, 86.09±3.99% and 90.23±3.05% for the three TF-CWs, B, C and D, respectively. Significant differences existed between the control(A) and the TF-CWs(B, C and D; P<0.05). While scenario D was the most efficient for NH4+-N removal, but A showed higher removal efficiency for NO3--N. The difference of TP removal rate were not significant. In general, we found that the idle/reaction time does not affect the removal rate of TOC. The NH4+-N removal rate was maximally efficient when the reaction depth ranged from 0 to 15 cm, and with increasing depth the removal rate slowed. This stage also showed rapidly rising NO3--N. With the increase of processing depth, TP concentration gradually reduced. Whereas TOCwas lower(0~20 mg/L) and further decreased with depth.2. The average intensity of nitrification were remarkable in four different simulators(P<0.05). Significant difference exist between A and the other three types. The largest average nitrification intensity appeared on D. The average denitrification intensity otherness in four simulators was also outstanding(P<0.05). The denitrification intensity was largest in continuous flow constructed wetland(A). Nitrification intensity of TF-CW matrix was significantly positive correlated with NH4+-N removal rate(R2=0.8497,P<0.05), while denitrification intensity and NO3--N effluent concentration showed a significantly negative correlation relationship(R2=0.8448,P<0.05). The nitrification intensity attained maximum where reaction depth ranges from 0cm to 30 cm and the maximum of denitrification intensity appeared on 30-60 cm.3. Through RDA we can konw the main factors affecting the removal of TN included DO(Dissoved Oxygen), RAT(idle/response time), ORP(oxidation reduction potential), and TOC. The main influencing factors of NH4+-N removal rate contain DO, RAT, ORP and Depth(processing depth), for NO3--N removal rate, the main factors are Cond(conductivity), Temp(water temperature), Sal(salinity) and p H. The main influencing factors of TP removal covered DO, RAT, Time and Depth. So choose the main influence factor of the every index as input layer when using BP neural network to simulate the water pollutant effluent concentration of TF-CW. Took pollution index concentrations as the output layer. Through trial and error way can got the numbers of node in hidden layer are 9, 11,12 and 9 respectively for TN, NH4+-N, NO3--N and TP. According the BP neural network training, results show that BP neural network model can effectively predict the effluent concentration of pollutants, model predicted value and actual value has certain correlation, but smaller range of error also exist. The fitting capacities of all indexes also can be obtained. Best fitting ability of TP effluent concentration, was 0.90076. For TN, NH4+-Nand NO3--N, are 0.67086、0.72854 and 0.69293.