There is a close relation between the structure and function of lake wetland ecosystems and its water level change. In recent years, due to intensive human activities and climate change, Poyang Lake has been subjected to emerging problems, such as earlier onset and extended duration of low water level condition,wetland degradation and sharp reduction of biodiversity and so on. The purpose of this paper is to explore the suitable water level that lake control project regulates. The CoupModel was used to simulate the respiration for maintenance and growth of Carex cinerascens from September to November during 1990-2009,and the relationship between the respiration amount of the Carex cinerascens and the dairy water level of the Xingzi hydrological station in Poyang Lake was established, using the artificial neural network,the response of the Poyang Lake wetland on the water level change was analyzed. The results indicate that the accumulated respiration amount of Carex presents a downward trend, which is 3.5% less after the Three Gorges Reservoir water storage than that before the Three Gorges Reservoir water storage. This study proposed three kinds of water level regulation schemes(high,medium and low water level), and the established neural network model was used to predict the variation of the accumulated respiration amount of Carex during the growth period under different regulation schemes. The results show that medium water level scheme improves the respiration amount of Carex after the Three Gorges Reservoir water storage obviously,which has obvious effect on the protection and restoration of the wetland ecosystem, while both other regulation schemes show no obvious effect. The results can provide a scientific basis for development of suitable regulation schemes of ecological water level in Poyang Lake wetland.