Railway inspection is an important task in railway maintenance to ensure safety. The fastener is a major part of the railway which fastens the tracks to the ground. The current article presents an efficient method to detect fasteners on the basis of image processing and pattern recognition techniques, which can be used to detect the absence of fasteners on the corresponding track in high-speed(up to 400 km/h). The Direction Field is extracted as the feature descriptor for recognition. In addition, the appropriate weight coefficient matrix is presented for robust and rapid matching in a complex environment. Experimental results are presented to show that the proposed method is computation efficient and robust for the detection of fasteners in a complex environment. Through the practical device fixed on the track inspection train, enough fastener samples are obtained, and the feasibility of the method is verified at 400 km/h.

Magnetostrictive linear position sensors (MLPS) are high-precision sensors used in the industrial field for measuring the propagation time of ultrasonic signals in a waveguide. To date, MLPS have attracted widespread attention for their accuracy, reliability, and cost-efficiency in performing non-contact, multiple measurements. However, the sensor, with its traditional structure, is susceptible to electromagnetic interference, which affects accuracy. In the present study, we propose a novel structure of MLPS that relies on two differential waveguides to improve the signal-to-noise ratio, common-mode rejection ratio, and accuracy of MLPS. The proposed sensor model can depict sensor performance and the relationship of sensor parameters. Experimental results with the new sensor indicate that the new structure can improve accuracy to ±0.1 mm higher than ±0.2 mm with a traditional structure. In addition, the proposed sensor shows a considerable improvement in temperature characteristics.

A novel three-point method using a grating eddy current absolute position sensor (GECS) for bridge deflection estimation is proposed in this paper. Real spatial positions of the measuring points along the span axis are directly used as relative reference points of each other rather than using any other auxiliary static reference points for measuring devices in a conventional method. Every three adjacent measuring points are defined as a measuring unit and a straight connecting bar with a GECS fixed on the center section of it links the two endpoints. In each measuring unit, the displacement of the mid-measuring point relative to the connecting bar measured by the GECS is defined as the relative deflection. Absolute deflections of each measuring point can be calculated from the relative deflections of all the measuring units directly without any correcting approaches. Principles of the three-point method and displacement measurement of the GECS are introduced in detail. Both static and dynamic experiments have been carried out on a simple beam bridge model, which demonstrate that the three-point deflection estimation method using the GECS is effective and offers a reliable way for bridge deflection estimation, especially for long-term monitoring.