Today, People's living environment in the rapid development of the industry has become a problem that can not be ignored. In order to alleviate the current situation of water pollution, this paper will make a research on it. The artificial neural network model is established based on the water quality characteristics of the domestic sewage. According to the structure of the model, the experimental device is designed to determine the weight of indicators in domestic sewage, and the experimental study is carried out. By comparing the simulated values and predicted values of 6 groups, then there is the following conclusion: through the system filter, the water quality of the effluent can reach the standard (GB18918-2002). A further analysis of the results, the conclusion that the sequencing batch biofilm reactor of the artificial neural network has a better performance in the purification of domestic sewage can be got. In order to guarantee the stability of the system, the performance is evaluated. Then there is the following conclusion: the simulation performance of the system is within the standard error range. The research of this paper has laid a foundation for the development of domestic sewage purification technology.