Point Rail of High-Speed Railway Turnout Induced Feature Extraction Using Oscillatory Behavior-Based Signal Sparse Decomposition

Turnouts are prone to failure due to their special structure. Wheel-rail force signals are informative about health conditions of turnouts; thus, the force signal analysis is a promising tool for assessing turnout health conditions. One of the key steps for assessing turnout health conditions is to identify the turnout-generated signal segment in measured force signals since there is always a deviation between the location of turnout-generated force signal in the measured force signal and the location of turnout recorded. Considering that the wheel-rail force impact caused by the point rail of turnouts is sparse, this article proposes a point rail-generated feature extraction method based on oscillatory behavior-based signal sparse decomposition (OBSSD). Due to the sparse and low oscillatory characteristics of the point rail-generated signal segment, the OBSSD can effectively extract the point rail-generated features from a blurred force signal. The OBSSD adopts a morphological component analysis (MCA) to map the analyzed signal into low and high oscillatory wavelets by constructing the basis via Tunable Q-factor wavelet transform (TQWT). The decomposition result depends on the selection of parameters involved; therefore, systematic discussion of parameter selections for the wheel-rail force signal processing is given and a convergence index (CI) for the parameter selection is presented. Then, by identifying the features caused by the point rail, the position of the turnout induced force signal segment in the wheel-rail force signals can be located, paving the way for the following turnout health condition assessment. The effectiveness of the proposed method is verified using measured wheel-rail force signals.