Security Region Estimation of the Peak of Track Irregularity Based on Dangerous Points Distribution Ratio and SVM

摘要

The security regions of track irregularities' peak amplitudes for highspeed railways are estimated by the simulation data and based on the theory of data classification, and the security regions boundaries with different speeds are solved, and this method can provide reference for the management and developing standards of track geometry on high-speed railways. First, by establishing a high-speed passenger vehicle dynamics simulation model using SIMPACK, inputs including vertical profile irregularities and alignment irregularities and outputs including three safety indicators data (derailment coefficient, wheel load reduction rate and wheel-rail lateral force) were collected under two different level track regularity states, and the irregularity data were marked as 'safe' or 'dangerous' according to certain safety evaluation criteria. And then, considering the difficulty of directly using SVM for irregularity data classification, a new classification method based on dangerous points distribution ratio combined with SVM was proposed which can reduce the difficulty and improve the efficiency greatly. At last, the test results and the comparisons of the domestic and international management standards of track geometry indicated that the proposed approach is effective and practicable.