The calculation of high-speed train speed based on multi-time-window speed-scanning method

To improve the accuracy of detecting the inter-spot speed of trains with train vibration waveform, this paper proposes the method of multi-time-window speed-scanning based on energy superposition, and verifies the accuracy and stability of this method with synthetic data and measured data. We recorded train vibration waves with 8 seismographs installed along the railway. We selected 100 waveforms at the same time points from continuous recordings of every seismograph. We analyzed these waveforms with the proposed method and the conventional cross-correlation algorithm to derive the train speeds, and compared the speeds to the actual speeds of the local trains. Our results show that the measurements of train speeds with the proposed method are stable and have no outliers, and the measured speeds are all agree with the real ones. To ensure this measurement effect, the distance between two seismographs at the ends should be longer than 290 meters; when the distance of two seismographs is between 290 and 365 meters, we recommend the use of more seismographs; when the distance is longer than 365 meters, the minimum number of seismographs is two. The proposed method is effective for both high-speed trains and normal trains, and it only requires a small land area for an easy installation of seismographs, which enhance the feasibility to use seismographs for practical train monitoring.