Vibrations are always present in mechanical
systems, and the railway infrastructure and rolling stock are no exception.
While large amounts of vibrations are commonly viewed as a source of irritation
or even discomfort, it can also be used as an indication of component damage
within a mechanical system. Vibration-based condition monitoring is an established
method for tracking the structural health of bridges, buildings, and conventional
machinery. They are now increasingly used in modern railway networks.
A team from the SMRT-NTU Corp Lab is currently
developing a MEMS-based wireless sensor network to provide vibration-based real-time
condition monitoring to both mobile and static railway assets. These can range
from monitoring infrastructures such as escalators and sliding screen doors, to
the operationally critical rolling stock and railway track. Irregularities in
the vibrational signatures can be picked up by the sensor network, and be used
as an indication of damage. The sensor network is envisioned to be a solution for
enabling predictive maintenance of railway assets.

The optimization of the parameters of the
wireless sensor network is a recurring theme in this research endeavor. Cost is
one of the major factors influencing the design of this sensor network. With
plans to deploy hundreds of nodes in a wide area, steps must be taken to ensure
that the final deployment is not cost-prohibitive. An example is the use of
MEMS technology to provide low-cost accelerometer nodes which collect data for
diagnostic information. In an increasingly data-driven industry, wireless sensor networks have the potential to
further progress the field of real-time condition monitoring. The team hopes
that this initiative will allow for the optimal planning of maintenance and the
minimization of operational delays.
