Prediction on building vibration induced by moving train based on support vector machine and wavelet analysis

Building vibration prediction induced by moving train can indicate the degree of vibration influence by moving train. With the prediction results, the corresponding measurements can be used to reduce the influence of vibration. To obtain an accurate prediction result, support vector machine (SVM) is used in this paper. Since the error in the recorded data affects the prediction performance of SVM obviously, wavelet analysis is adopted to filter the input data. The prediction model based on SVM and wavelet analysis is validated by the data of field experiments. The results show that the prediction model can provide a good performance compared with the measured values.