It is very important to timely and accurately identify the water behavior of residents for developing effective household water demand management strategy,enhancing water security and improving the current planning and management of water infrastructure.Aiming at the water flow characteristics each event, the paper proposed two recognition methods of residential water behavior through the analysis of several typical residential water behaviors. It established the model of behavior recognition of different types of residents by using HMM and BP neural network and achieved the real-time effective recognition of residential water behavior.These two methods were extracted from the training set of water flow characteristics of different behavior and a reference model of the behavior was established.Then the water flow characteristics extracted from the set of test were matched to the reference model and obtained the recognition results.Finally,it compared and analyzed the results of two identification methods.The result concluded that under the water event of different traffic patterns,the identification accuracy of HMM is about 6% higher than that of BP neural network model.Under the water event of similar traffic patterns,the identification accuracy of BP neural network model is about 6% higher that of HMM.