Eutrophication of surface water bodies resulted from pollution of the farmland drainage water is a serious water pollution problem recently, which has been paid more and more attention. Phytoremediation is to use specific aquatic or terrestrial plants for water pollutants such as nitrogen in situ absorption, degradation and fixing, with low cost and good effect of removing nitrogen pollutants, which is gradually used in sewage disposal. However, impact factor of phytoremediation is not known comprehensively enough since people have done research on it recently, and study on related dynamics is not much either. So many numerical models are not accurate enough, and the selection of reaction parameters is not unified. In this paper, two kinds of phytoremediation constructed wetlands such as surface wetland named Cell A and vertical flow constructed wetland named Cell B are used. Through experiments design, operation of the field and sampling analysis, combined with the principle of nitrogen transformation model, phytoremediation of farmland in the water cycle nitrogen ecology dynamic conceptual numerical model is developed. A kinetic model of nitrogen physical-chemical biodegradation reaction is established in this study. And genetic algorithm with the least-squares method of nonlinear optimization is used via MATLAB, which is an independent program to determine the unknown parameters in the developed model. Simulated results showed that the concentration of ammonia nitrogen, nitrate nitrogen and organic nitrogen in the effluent of Cell A is better than other pools. The ecological dynamic model of nitrogen cycle and the first-order kinetic model of removing total nitrogen (TN) were compared and analyzed in the simulation of the Cell A. It is indicated that the developed ecological dynamic model in this study could accurately predict the change of the effluent concentration in the constructed wetland. And the model in this study can provide theoretical support for the design and operation of the constructed wetland and the prediction of the effluent concentration to facilitate the mastery for the actual purification effect of the constructed wetland.