Currently the health of a locomotive engine is analyzed by taking oil samples from the engine only when it comes to the pit. These samples are then sent to labs for analyses and a manual probing is required to interpret the health of the engine. There is always a delay of 15 days associated with the above process which leads a recurring road failures in the freight transporting system. What a customer would wish to have a system that alters them before a road failure. Is that a possibility with the current set-up for about 25000 locomotives running across the globe. Hey, here is a good news.. we have a solution for that.

What it does

The solution is to get the instantaneous analog data from the locomotive, analyze using physics and alert it before it reaches the pit. Our demo, transmits the existing analog data to the cloud using two approaches. ( a ) Through a R-Pi ( b ) Through a NI data logger. This data is pushed to the cloud using Python based websocket approach that we have developed. This data is used by the physics based micro services that we have built in PREDIX for to predict other engine health parameters such as TBN, Soot & Viscosity. There are limits for these parameters and a smart alerting system is built in the code which alerts the customer through a UI. An email or SMS alerts can be further included in the package.

How we built it

We will tell you while presenting... Sorry do not have time

Challenges we ran into

The objective is to push the time series data to the cloud first. PREDIX machine was only supporting Modbus. We made it to work using R-pi, Web socket client using python WSS library. We had to understand how the WSS authorization and Json building which we have done for the first time

Accomplishments that we're proud of

We are able to live stream the sensor data to the cloud. This can solve all other problems related to maintenance of a locomotive

What we learned

  1. How to push data from sensor to cloud
  2. Run a micro service linking the sensor data to do calculation
  3. etc... ## What's next for Oil Health Advisor for GE products We have trademarked a part of this concept already and planning sell this as a software solution.

Built With

  • angular-html
  • dht22-temerature-&-humidity-sensor
  • java
  • k-type-thermocouple
  • modbus-simulator
  • ni-data-logger
  • postgress-db
  • predix
  • predix-assets
  • python
  • rasberry-pi-3
  • scipy
  • time-series
  • websocket-client
Share this project: