Automated Failure Detection Public Transit Company

Case Study

BACKGROUND

CLIENT - ELECTRIC TRAIN OPERATOR

Our  client is a North American train and bus operator. Their focus is safety,  train performance, and cost optimization.

They came to Machinery Analytics with the goal to maximize their efficiency through quicker and more accurate train failure identification. Identifying failures in a timely matter would allow their team to more quickly and efficiently address them.

 

We have so far created an algorithm that can identify abnormal events up to days/hours before maintenance events, without any prior knowledge of the system.

Instagram Posts (33).png

We started by processing 50GB or 100M data rows of maintenance and operational data.


This image shows a very small section of the data.
As you can see, it's very complex and time-varying.

STARTING DATASET

We were able to use the data given to develop an algorithm that achieved the following:

identified system failures up to 160 hours before the maintenance event occured

with no prior knowledge of train system

OUR MODEL

 

RESULTS

looked at

8 distinct trains

able to identify

3 failure types

able to detect failures

160 hours before event occurred

Are you facing similar obstacles in your organization?

At Machinery Analytics, we believe that skilled professionals should always have the best tools for the job. If you're ready to save time, money, and to improve product safety, contact us for a demo or to book a call!