Global climate change and ocean warming have increased the frequency of extreme weather events in East Asia’s coastal regions, yet the lack of long-term tide level records hinders the understanding of storm surge evolution and disaster preparedness. This study aims to develop a machine learning–based method to efficiently identify historical typhoons impacting the Yangtze Estuary and enable streamlined numerical storm surge simulations. Using a random forest model trained with observed tide level data from 1989 to 2020 and several typhoon parameters, historical typhoons from 1949 to 1988 were classified for their potential to induce storm surges. The Regional Historical Storm Surge Simulation toolbox (RHSS-toolbox) was also introduced, integrating optimal typhoon tracks and reanalysis wind data to automate the creation of driving wind fields and support batch simulations with the FVCOM model. The results indicate that 90 typhoons were correctly identified, with a recognition accuracy of 96.6%. No false negatives were observed relative to the baseline ISYE-R500 method, and the simulation workload was reduced by 93% compared with traditional methods. The RHSS-toolbox improved nearshore wind field accuracy and enhanced the calculation efficiency for regional historical storm surge simulations.. In conclusion, this framework offers a robust and efficient solution for historical storm surge research in data-scarce regions and provides significant scientific support for coastal disaster prevention and long-term planning.