Railway is an important infrastructure, and tracks loaded with frequent, heavy loads are increasingly vulnerable to flaws in the rails. Therefore, routine inspections of rail tracks are required to prevent a catastrophic scenario from happening. Most of the current inspection techniques have limited volumetric inspection coverage, hence they can only detect local defects in a point-by-point type of inspection fashion. Use of these techniques is very time-consuming and ineffective.

Ultrasonic guided wave (UGW) is an effective technique for inspecting the health condition of rail tracks. It can be used to inspect the integrity of a specimen from a single point, as the emitted UGW can travel inside a rail over a longer distance. Therefore, UGW can be used to inspect rails in a line-to-line fashion. Another challenge with rail track inspection is that it is not possible to mount a physical sensor on a rail. Hence, the concept of a transduction system consists of a non-contact, 3D type of laser system that can be tailored to emit and receive UGW is promising. Moreover, using a 3D laser scanner to receive the emitted UGW signals does not suffer significantly from attenuation like that received by air-coupled transducers. It is because their efficiencies are affected by the air gap, the air density and velocity, as well as the turbulence at the time of sensing. Therefore, the use of laser to emit a desired UGW mode to a specimen and the use of a laser 3D scanner to receive the defect-reflected UGW signal have become popular. Nonetheless, despite possessing long-range excitation capability and being less influenced by air, there are several disadvantages of using laser to emit and receive UGW signals. One disadvantage is laser can only emit UGW signal in a wideband type of frequency range. That is, the emission of laser-generated UGW often generates multiple wave modes simultaneously. The second disadvantage is its low excitation energy which may lead to a low Signal-to-Noise Ratio (SNR). Although the laser-generated UGW can be emitted at a very high energy than that emitted by other conventional UGW-based sensors, too strong an excitation energy may damage a rail surface.

To solve the wideband problem, the use of an optical device to modulate the emitted laser-generated UGW to a narrowband frequency range is preferred. A device developed recently, called Integrated Optical Mach–Zehnder system (IOMZ), was designed to limit the frequency range of emitted laser-generated UGW to a desired narrowband range. However, IOMZ is prone to the ease of having misalignment. Misaligned IOMZ may lead to a curved laser line array pattern (LAP) and/or letting LAP to have different line widths. These two issues have a significant negative impact to the laser-generated UGW signals. An improved version of IOMZ was developed to address these problems and reported here. The improved version is more compact and does not generate LAP with different line widths. It is also built with less optics compared to that of the original IOMZ. The improved IOMZ is more compact, contains less optics, and incorporates a different optical path that can prevent the generation of LAP with different line widths. With the help of improved IMOZ, it is more easily to be aligned to generate UGW in a desired frequency range. However, both IOMZ and improved IOMZ have identical optical configurations that govern the LAP generation process. Both of them require precise alignment to form equidistant LAP. The optical path difference (OPD) is one of the key features that affect the LAP generation process. If the OPD is not equal to zero, this will result in curved LAP. Consequently, the laser energy cannot be focused effectively to excite the desired UGW mode. To overcome this problem, another new design, called Sagnac Interferometer-based Optical System (SIOS), was developed. From the experimental results, they prove that SIOS can outperform IOMZ and the improved IOMZ. SIOS has a completely new optical configuration which makes use of the design based on common-path interferometry. All laser beams share the same optical path. If any of the optical components was shifted slightly, the optical path would remain the same and the optical path difference would always be zero. Hence, the generation of equidistant LAP is unaffected by changing the optical path. As a result, the laser-generated UGW is always within the desired frequency range. Adjustment of LAP on SIOS is controlled by one optical lens only. Thus, it provides more flexibility and relatively ease in tuning the operation frequency of the laser-generated UGW. Furthermore, SIOS has a transmission efficiency of up to 86%, which is much higher than that offered by the conventional physical slit mask and shadow mask reported by other researchers used to change the frequency band of the emitted UGW.

The allowed strength of laser-generated UGW is not strong enough to reveal rail subsurface defects. Such defect reflection cannot be easily revealed in a B-scan plot nor in a A-scan temporal plot. To enhance its ability in revealing defects, a signal processing technique, called chaotic oscillator, was integrated with the 3D laser system to reveal weak signals reflected by subsurface defects. The integrated system is called noncontact laser-based duffing oscillator system (NLDOS). One of the key features of NLDOS is its instability. For example, adding a weak UGW signal to the system may lead to a dramatic change in the system’s response because it creates instability in the system. A weak UGW signal can be revealed by analysing the response of NLDOS. Since the performance of NLDOS is not affected by the presence of strong random noise, it can reliably identify all received UGW signals that are reflected by having a surface, a subsurface and an internal defect from strong noise. Based on the above discussions, the contributions generated by this research work are listed as follows. First, the feasibility and capability of using a completely laser-based transduction system to emit a desired UGW mode and the use of a noncontact 3D laser scanner to receive the UGW signals that had travelled inside a rail were revealed here the first time. Second, the design of a purely optical lens-formed SIOS that can replace conventional metal masks to emit narrowband UGW that can substantially enhance the function of the original IOMZ was successfully tested and verified. Third, the design of a new signal processing algorithms, the NLDOS, that can reveal weak UGW signals reflected by both surface and subsurface defects from strong noise was also successfully tested and verified. With the help of such complete 3D laser-based transduction system to emit and receive UGW in narrowband by SIOS and then analysed the reflected UGEW signals by the advanced chaotic oscillator, both surface and subsurface rail defects can be detected in a more efficient line-to-line fashion. Hence, by using the above new designs to inspect the integrity of rails, the safety of trains running on the rails substantially enhanced and the operation and maintenance of rails can be significantly improved.