To achieve real-time monitoring data collection, intelligent evaluation and warning of safety risks, and rapid disposal of emergency plans for steel cofferdam structures during the construction process of deep water foundations of bridges. Research on intelligent safety control of steel cofferdams of deep water foundations of bridges was conducted based on digital twins. Firstly, BIM software was used to construct a virtual model of the entire construction process of the steel cofferdam, and IoT devices were used to collect real-time data on the stress, displacement, and construction water level of the steel cofferdam structure. Secondly, by using the Web-GL engine to couple physical data with virtual models, a real-time interconnected digital twin scene was built. Finally, based on finite element simulation analysis and construction technology, a structural safety assessment index system was established, as well as a safety warning and response system, achieving intelligent safety control throughout the entire process of steel cofferdam construction. The reliability and scientificity of this technology have been verified through practical engineering applications, and the research results will effectively improve the efficiency and accuracy of steel cofferdam monitoring, ensuring controllable construction safety risks.