Based on the maximum flux principle(MFP),a water quality evaluation model for surface water ecosystem is presented by using self-organization map(SOM)neural network simulation algorithm from the aspect of systematic structural evolution.This evaluation model is applied to the case of surface water ecosystem in Xindu District of Chengdu City in China.The values reflecting the water quality of five cross-sections of the system at different developing stages are obtained,with stable values of 1.438,2.952,1.869,2.443 and 2.479,respectively.The simulation also indicates that the larger the value,the more serious the water pollution.Furthermore,a classification graph is given to reflect the evolution of structural pattern.The combination of MFP and SOM neural network reveals the formation of different structural patterns in the system during the interaction of internal components.It is shown that a dominant pattern is finally reserved,which starts from a variety of combination patterns for a time period.The results agree with those from traditional evaluation methods,which indicates that the proposed model has high accuracy.This model embodies the evolutionary dynamic mechanisms and characteristics of temporal and spatial changes,which helps to guide the prediction of water quality status of surface water ecosystem.