Abstract: The excavation of a foundation pit changes the stress state of the supporting structure and surrounding rock and soil, so it is necessary to monitor the foundation pit continuously in engineering construction. In order to solve the problems in the existing system, this paper developed a foundation pit early warning system using various technologies, such as the B/S mode, Vue front-end, C^# object-oriented language, SQLServer 2012 database, and LSTM neural network algorithm model based on the Python language. The system realizes the functions of centralized information management, data storage and viewing, automatic calculation of data algorithm, automatic drawing of charts, automatic alarm and early warning, rapid generation of alarm reports, and so on. The application in the excavation process of a foundation pit in Suzhou proved that the system can provide technical support and guarantee for the construction departments to accurately get the deformation of the supporting structure and surrounding soil using two early warning methods: monitoring early warning and advance prediction early warning based on the LSTM neural network model.