Nobody likes their train journey to be disrupted, trains run on wheels... and wheels wear down. Currently if a wheel breaks, the train is taken out of service for a whole day inconveniencing customers and costing train companies thousands.

However we found a way to repurpose large amounts of maintenance data to predict premature wheel failure so they can be serviced in groups before they break down. Currently one wheel set gets changed in a maintenance window which is highly inefficient and can lead to maintenance clustering.

WheelX is a data prediction platform that agregates wheel data, calculating the chance of failure and alerts staff to allow them to take preemptive actions.

What it does

Predicts when a train needs it's wheels changed prematurely. Combines multiple data sources and calculates which trains are likely to encounter problems before their next maintenance window. This prevents the need to recall trains which have routes planned for them thus minimising disruptions.

How we built it

Using machine learning and statistical data analysis. See tags.

Challenges we ran into

Finishing this story on time. The lack of internet connectivity particularly plagued us as we were working on several different parts, each outside or particular comfort zones. This was made more difficult as we were not able to find easy solutions to our problems using the internet. This forced to help each other and collaborate in pairs to complete each part. However these challenges seems to bring us together and produce a solution that would not be possible by any one of us individually (even with the 4G)

Accomplishments that we're proud of

We solved the challenge. After completing our prototype, we pitched it to the mentors from Bombardier and they expressed their happiness with the solution we have provided. Thus verifying out idea by the people who set the challenge.

What we learned

During this task, we worked closely with the people who created the challenge.

Built With

Share this project: