5. Outlooks of SHM on HSR

With embedded hybrid monitoring systems of FBG and PZT sensors, the SHM techniques have shown their promising prospect in HSR, enabling real-time monitoring of structural conditions of in-service trains and rail infrastructure. To realise large-scale utilisation on numerous HSR lines worldwide, practical solutions ought to be achieved concerning both economic and efficient aspects, answering for the need of early warning and quick decision-making upon emergencies in high-speed operation and guiding the potential development direction of SHM applications on HSR in the coming decades.

Wireless sensing network (WSN) provides a cost-effective approach eliminating wires and enabling remote sensing, which largely enhances the practical applicability of SHM [80, 81]. A wireless-based system was designed to monitor the performance of rail vehicles by Nejikovsky and Keller [82]. The communication in the WSN system can be made through satellite and Ethernet, while data are uploaded onto cloud for storage and transmission to control room far away from site; system data transmission plan can be found in the aforementioned railway tunnel deformation project [76]. Particularly, in terms of near-field communication, radio frequency identification (RFID) has been proposed as a competitive candidate [83], which provides a new thinking on emerging RFID modules in normal sensors. The passive RFID sensors embedded in the HSR structures need no wired power supply and can be activated by passing trains, sending structural condition information.

Continuous online monitoring of HSR over multiple HSR lines puts forward the difficulty in storage and analysis with massive data collected. The authors' team has long been dedicated to damage diagnosis and prognosis of HSR based on monitoring data with updating and learning methods. Facing the data amount issue, compressive sensing, which is able to sample data at sub-Nyquist sampling rate while maintaining almost all the original information, is being actively investigated to streamline the axle box acceleration data from an operating high-speed train and has successfully verified the feasibility of sub-Nyquist data acquisition in HSR online monitoring [84, 85]. This is of great significance to wireless sensing and RFID where transmitted data amount is limited.
