A construction and analysis system of touch screen user keypress behavior pattern, and an identity recognition method thereof. Data analysis is performed by using historical keypress information of inputting a password by using a soft keyboard, a corresponding neural network model is established and model calculation is performed to new to-be-detected data to recognize a user identity; the system consists of a user data acquisition module, a data preprocessing module, a model training module and a user identity authentication module; the user data acquisition module is responsible for acquiring time sequence information, pressure and contact area information; the data preprocessing module is responsible for acquiring data, removing dirty data and normalizing to the data; the model training module is responsible for analysing input patterns and establishing models; and the user identity authentication module is for performing model calculation to new to-be-detected data to recognize user identities and improve the security of user account passwords.