Predictive maintenance through combining data

Mani gave a workshop on using data to better predict maintenance of equipment and infrastructure of public transportation, such as streetcar or train transportation. According to Mani, it is important to establish relationships between different data sets such as those of infrastructure and vehicles for this purpose. "It's about the predictive value of data. The data source of one, can offer insights into the operation of another. We should therefore share more data within the industry."

Prevention rather than cure

"If you combine data, the carrier can see, for example, that an additional capacity request is coming and can adjust its deployment accordingly." Based on such analyses, the move can also be made to preventive maintenance. "Now we do reactive maintenance," says Mani. "You bring in a subway when it's needed. But we would rather prevent than cure." Data can help with this. "Smart maintenance is ultimately about seeing problems coming. The better we can assess that, the more cost-effective maintenance can become."

Know what you're measuring

The key here is knowing what is being measured and what can then be done with the data. Mani takes as an example an incident on Jan. 15, 2014 near Hilversum. A train derailed because a switch flipped while a train was passing over it. InTraffic, in collaboration with ProRail, looked at 7,000 other switches following the incident at Hilversum and found a number of switches where the same thing was likely to happen.

InTraffic then developed Landelijke Infra Monitoring (LIM) together with ProRail. This is used by the rail operator to smartly schedule switch maintenance to prevent breakdowns.

Combining data for predictive maintenance

For this to happen, it is important to start analyzing and combining data sets together, Mani believes. "By combining data from all kinds of sources, you can extract such interesting insights."

Read the full interview with Jillis Manis on spoorpro.