PSI - Issue 4

Uwe Oßberger et al. / Procedia Structural Integrity 4 (2017) 106–114 Author name / Structural Integrity Procedia 00 (2017) 000 – 000

110

5

Fig. 4 Signal processing chain to compare strain signals of a frequently passing train: (a) Strain signals of 6 different ÖBB 4024 trains with 10 axles; (b) Transformed strain signals of a; the influences of different velocities/accelerations of the passing trains as well as the start time have been eliminated via dynamic time warping (c) top: consensus curve of the 6 ÖBB 4024 trains, bottom: 50% confidence interval for the consensus curve, details published by Kollment et al. (2015). Note that the corresponding signal processing on the strain data goes well beyond a mere threshold analysis on the strain maxima but covers the analysis of the curve shape through the evaluation of confidence envelope curves. The consensus curve can be used as a fingerprint for a "ÖBB 4024 train passing a healthy crossing nose". Correspondingly, strain signals deriving from new ÖBB 4024 passings can be compared with the initial "healthy crossing" data. It stands to reason that after this procedure the detected and quantified changes in the measured strain data can be attributed to changes of the system "wheel passing a crossing nose". The mathematical algorithms behind the developed signal processing chain can be found in the corresponding publication by Kollment et al. (2016). The signal processing chain shall serve as base for continuous monitoring of fixed point crossing nose via strain measurements. To this end the following work flow is proposed:  Record the strain signals on the crossing continuously,  identify train type, direction and speed,  pick trains of a specific type and direction (e.g. a specific train/traction unit that passes the in regular intervals),

 point out trains with irregular wheels (see Kollment et al., 2016),  point out changes in residual strain (see Kollment et al., 2016),

 exclude all axles except the drivetrain axles (to exclude the influence of different axle loadings) and  compare the strain data with all preceding transitions of the same train and monitor the changes.

Based on the assumptions above it can be concluded that strain signals of driven axles of the ÖBB 4024 train adjusted to the actual train speed the remaining influencing parameters are the crossing nose geometry and the bedding. To distinguish between the geometry changes and the bedding and to correlate the observed signal change with the crossing nose geometry change it was therefore necessary to quantify the geometry changes of the crossing nose with load time.

3. Geometry measurement and quantification of changes

To quantify the geometry changes of the new tool steel crossing nose due to wear and plastic deformation in service a laser based non-contacting measurement of the crossing geometry has been set up.