Abstract. Liquefaction-induced hazards are responsible for considerable damages to engineering structures during major earthquakes. Presently, there is not any effective empirical approach that can assess different liquefaction-induced hazards in one model, such as sand boils, ground cracks, settlement, and lateral spreading, due to the uncertainties and complexity of multiple related factors of seismic liquefaction and liquefaction-induced hazards. This study used Bayesian network method to integrate multiple important factors of seismic liquefaction, sand boils, ground cracks, settlement and lateral spreading into a model based on standard penetration test historical data, so that the constructed Bayesian network model can assess the four different liquefaction-induced hazards together for free fields. In the study case, compared with the artificial neural network technology and the Ishihara and Yoshimine simplified method, the Bayesian network method performed a better classification ability, because its prediction probabilities of Accuracy, Brier score, Recall, Precision, and area under the curve of receiver operating characteristic (AUC of ROC) are better, which illustrated that the Bayesian network method is a good alternative tool for risk assessment of liquefaction-induced hazards. Furthermore, the performances of the application of the BN model in estimating liquefaction-induced hazards in the Japan's Northeast Pacific Offshore Earthquake also prove the correctness and reliability of it compared with the liquefaction potential index approach. Except for assessing the severity of hazards induced by soil liquefaction, the proposed Bayesian network model can also predict whether the soil is liquefied or not after an earthquake, and it can deduce the process of a chain reaction of the liquefaction-induced hazards and do backward reasoning, the assessment results from the proposed model could provide informative guidelines for decision-makers to detect damage state of a field induced by liquefaction.
Information Security Risk Assessment of Smartphones Using Bayesian Networks
