This paper is a technical report on the preliminary work for a track fault detection system to be used on standard trains using low-cost sensor technology. The focus is on the system description as well as the presentation of a first approach for data evaluation. The overall objective of the study is the subsequent detection and localization of irregularities of a track which occur due to discontinuities on the rail or changes in the stiffness of the subsoil. Such discontinuities influence both the wavelength and the amplitude of the vertical acceleration signal. The influences occur e. g. when crossing road crossings, switches or bridges but also when passing track faults. These characteristic signal changes are to be used to detect track faults. To measure the signals, a single-axis and a triple-axis accelerometer are installed on the bogie of a standard train. In addition, a GNSS receiver is installed to ensure localization of the irregularity. A real-time computer is used to record and synchronize the data.
Although methods exist that are capable of identifying and classifying discontinuous points on the track, it was initially decided to work with a normalized cross-correlation that only serves to detect discontinuities. As more measurements become available, more robust analyses can be performed to classify these discontinuities. In the meantime, the normalized cross-correlation will be used to work with the available data.