MainTrain

Inspiration

Angel Trains and Eurostar's predictive maintenance challenges

What it does

Our product uses an array of sensors with on-train processing and machine learning to tell train operators when and how components on their train are about to fail. It then can automatically order parts and book its own slot in the depot to automate human tasks, saving time and money.

How we built it

We used a Particle Photon WiFi board, microphone and vibration motor for the hardware demo. For data analysis and visualisation we used golang and a javascript library. We used excel and sigview to analyse the existing datasets and train our model to then predict in real time how components would fail.

Challenges we ran into

Sensor connectivity, Using an immature platform for the hardware demo(limited library availability), lack of detailed financial knowledge around the issue, WiFi connectivity.

Accomplishments that we're proud of

Our branding and design, the interactiveness of our demo, the fact that we were the only hardware hack on our train, being able to use our individual skills to maximum effect to help the team.