English：

In deep foundation excavation construction for subway projects, dynamic changes in the groundwater level directly impact engineering safety, while traditional monitoring methods suffer from insufficient predictive accuracy and poor visualization interactivity. To address this, this paper proposes an intelligent monitoring method that combines time series analysis with dynamic visualization. Based on automated monitoring data from the South Chengshan Road Station on the Shaoxing Metro Line 2, ARIMA was employed for water level forecasting and utilized Vue and ECharts to build a realtime interactive visualization platform. The results indicate that the onestep prediction error of the ARIMA model is less than 2cm. The visualization system can display water level changes in real time and supports interactive data queries, significantly improving monitoring efficiency and resolving the lag problems associated with traditional water level monitoring.