Probabilistic prediction of metro induced ground-borne vibration and its accuracy evaluation

The environmental vibration caused by the metro trains has remarkable uncertainties due to some uncertain parameters in the source and propagation path of the vibration. Based on the complex principal component analysis, a probabilistic prediction approach is presented to quantify the uncertainty of the wheel-rail force, which can be used to predict the ground-borne vibration in the probabilistic framework. Furthermore, an evaluation index is proposed to quantify the accuracy of the predicted results. This index can be used to consider not only the bias between the predicted and measured results but also their respective uncertainties. A case study is made to validate the probabilistic prediction procedure and the evaluation of the prediction accuracy. The results show that the predicted and measured vibration levels have good agreement, and the evaluation index can effectively quantify the prediction accuracy at frequencies in one-third octave band. • A probabilistic method is proposed in frequency domain to quantify the train loads. • Probabilistic prediction of the metro induced ground-borne vibration is realized. • A distribution-oriented index is presented to evaluate the prediction accuracy.