3.2 Online wheel condition monitoring and defect detection

Existing online wheel condition monitoring systems mainly include trackside wheel impact load detector (WILD), force gauges installed on sleeper pads, distributed sensors based on Brillouin optical time domain analysis (BOTDA), accelerometer-based trackside detector, acoustic detectors, laser- and video camera-based detectors, etc.

## 3.2.1 Trackside WILD and wheel: rail interaction detector

By deploying strain gauges and accelerometers on the rail, it is possible to measure wheel-rail contact force or rail acceleration response when a train passes over the instrumented rail section. These devices report impact as either a force at the wheel-rail contact interface or a relative measure of the defect [10]. The most common WILD is composed of a series of strain gauge load circuits mounted on the neutral axis of the rail between two adjacent fasteners in several consecutive sleeper bays to quantify the wheel-rail interaction force.

Johansson and Nielsen [1] made use of this set-up to build a detector on the rail web in nine consecutive sleeper bays. Nielsen and Oscarsson [32] used both rail web and rail foot strain gauges to measure the wheel impact load and rail bending moment. Stratman et al. [33] proposes new criteria for removal of wheels with high likelihood of failure, based on two real-time SHM trends that were developed using data collected from in-service trains. Filograno et al. [34] developed an FBG-based sensing system comprising FBG strain gauges mounted at both rail web and rail foot enables train identification, axle counting, speed and acceleration detection, wheel imperfections monitoring and dynamic load calculation. They have expanded the application of this system in the Madrid-Barcelona HSR line [34]. An FBG-based wheel imperfection detection system that can provide in-service measurement of wheel condition was developed by The Hong Kong Polytechnic. It offers a comprehensive health monitoring scheme for vehicle and track in the entire railway network of Hong Kong [35]. A monitoring system has been proposed with FBG sensors implemented on rail tracks to detect wheel local defects such as wheel flats and polygonisation [36]. The impacts of wheel/rail interactions caused by wheel local defects are reflected as subtle anomalies in response to signals collected by FBG sensors, and the deployed system is shown in Figure 3a. The detecting results match well with those from offline inspection (Figure 3b).

In addition to the strain gauge-based detector, there are other methods for online wheel load measurement to assess the condition of passing wheels, such as force gauges installed on sleeper pads, distributed sensors based on Brillouin optical time domain analysis (BOTDA), etc. Besides, there are also some commercial WILDs, such as WheelChex® system, GOTCHA system (optic fibre-based wheelflat detection and axle load measurement system), and MULTIRAIL WheelScan.

## 3.2.2 Trackside rail acceleration and noise detector

Accelerometer-based systems can provide 100% coverage of the circumference of a wheel of any size in defect detection [10]. Skarlatos et al. [37] used two B&K accelerometers placed on the rail foot to pick up the rail vibration signals for diagnosis of wheel defects. Belotti et al. [38] used four consecutive accelerometers and an inductive axle-counter block which help to discriminate the response corresponding to each wheel. Seco et al. [39] proposed a trackside detector which has eight accelerometers installed on both bend zone and straight zone. However, the acceleration data are difficult to convert to wheel-rail impact load, which is widely used as wheel local defect indicator [5]. This is mainly because the measured acceleration signal could not directly refer to the excitation of each wheels, and

Figure 3.

High-speed train wheel defect detection using online FBG-based system [36]. (a) System configuration. (b) Detection results (left—online detection; right—offline inspection).

Contemporary Inspection and Monitoring for High-Speed Rail System DOI: http://dx.doi.org/10.5772/intechopen.81159

sometimes an additional axle counter is needed [23]. Furthermore, their performance might be limited by their repeatability and by the analysis applied to the accelerations acquired.

The commonly used noise detector is called trackside acoustic array detectors (TAADs), which make use of arrays of high-fidelity microphones to listen to the audible noises produced by the passing trains [40]. There are also some commercially available systems, such as trackside acoustic detection system (TADSTM) and the RailBAMTM [41, 42]. However, these systems are specialised in wheel bearing fault diagnosis rather than wheel tread defect detection. For the slight flat defect of high-speed train wheel (flat depth < 0.5 mm), this method may not be applicable because when the train runs at high speed (>200 km/h), the prediction accuracy can be limited [11].

## 3.2.3 Detection methods based on laser and video cameras

There are various types of wheel roughness monitoring systems based on laser and video cameras. Some typical or well-known detectors include wheel profile detectors (WPDs), MBV-systems, wheel profile measurement system (WPMS) and those based on light illumination devices, light-sensing devices, charge-coupled device (CCD) camera, and laser displacement sensors (LDSs).

WPDs are based on a combination of lasers and video cameras to automatically measure the wheel profile while train is in motion [40]. These data acquired from WPDs include wheel profile and wear, wheel diameter, height and thickness of the flange, back-to-back distance and wheel inclination. A prototype of a condition monitoring system called MBV-systems is presented by Lagnebäck [43] for measuring the profile of the wheels with a laser and a camera. An automatic WPMS based on laser and high-speed camera was installed on an iron ore line in Sweden in 2011 and can measure the wheel profile for speeds up to 140 km/h [44]. This system, which is in a CBM manner, has been attracting more and more interest from maintenance engineers in the Swedish railway sector. Zhang et al. [45] presented an online non-contact method for measuring a wheelset's geometric parameters based on the opto-electronic measuring technique. The system contains a charge-coupled device (CCD) camera with a selected optical lens and a frame grabber, which was used to capture the image of the light profile of the wheelset illuminated by a linear laser. Besides, there are some newly designed laser-based online detectors, which are immune to vibration and on-site noise, easy to calibrate, with high efficiency of data acquisition and with high accuracy of positioning [46, 47].
