Abstract: The safety status of tunnel construction workers is mainly judged through workers′ physical signs and cave environment data, and the early warning of the abnormal status usually requires professional staff to make a rapid judgment in a short time, which is costly and inefficient in operation and maintenance. To address this issue, a oneclass support vector machinebased personnel safety status detection and early warning model is proposed. First, data from the actual tunnel construction scenario are collected using sensor devices deployed in the field, and a oneclass support vector machine model for abnormal state prediction is built. Second, the model is retained for early warning state testing, and relevant environmental data as well as construction personnel physical signs data are collected from engineering examples. In addition, horizontal different parameter model experiments and vertical different early warning state proportional data experiments are conducted to evaluate the model′s performance for personnel information security state judgment. The experimental results show that the accuracy rate of personnel security status early warning reaches more than 90%.

Key words: one-class support vector machine, personnel security status detection, tunnel construction, one-class support vector machine model