Research on Bayesian Source Localization of Intrusion Events Based on Adaptive Selection of P-Wave First Arrivals

Distributed acoustic sensing (DAS) technology plays an increasingly important role in underground engineering safety monitoring where the identification of intrusion events and source localization are key issues. In response to the issues of poor location stability of single method-based P-wave arrival time data and inconsideration of P-wave arrival time data uncertainty in source localization, this study proposed an intrusion event Bayesian source localization method for distributed acoustic sensing that adapted the selection of P-wave arrival time data and considered its uncertainty. Specifically, we established a source localization objective function for intrusion events that attenuates with distance, developed a P-wave arrival time data adaptive selection and source preliminary localization method, and then proposed a Bayesian source localization method that considered the uncertainty of P-wave arrival time data, providing the source localization results and its uncertainty. Tests on intrusion event source localization in the drainage tunnel show that: (1) Sensor data closer to the intrusion event location have significantly greater localization weight, while the weights of sensor data from further distances tend to zero; (2) For sensors closer to the propagation distance, the adaptively selected P-wave arrival time is consistent with the theoretical propagation time; (3) The Bayesian source localization considering P-wave arrival time uncertainty converges quickly, with the source localization results showing a circular distribution of uncertainty in the horizontal direction and an elliptical distribution in the vertical direction, with an intrusion event localization error within 1 meter; (4) The average location errors of rock fall, water release, human walking, and manual knocking tests are 0.98 m, 0.81 m, 1.82 m and 0.35 m, respectively. In summary, this study provides a new approach to the adaptive selection of P-wave arrival time data and the analysis of uncertainty in source localization results.