To enhance the timeliness and accuracy of storm surge forecasting in the coastal areas of Ningbo City, an intelligent storm surge prediction system using a long short-term memory (LSTM) model was developed. Based on historical typhoon records and virtual typhoon information, storm surges at tidal stations along the coastal areas of Ningbo City during typhoon events were calculated with the ADCIRC hydrodynamic model, creating a storm surge sample database. The LSTM model was applied to train the sample data of storm surges in the coastal areas of Ningbo City, and a robust and efficient intelligent forecasting model for storm surges in the coastal areas of Ningbo City was established through sample optimization and parameter calibration. The test results of No. 202212 Typhoon Muifa show that when the number of trained samples exceeds 400, the constructed forecasting model can accurately achieve short-term (1-12 hours) forecasting of storm surges in the coastal area of Ningbo City. However, when the forecast period exceeds 12 hours, there may be significant deviations between the forecasting results and the actual data.