airplane engine

Inspirations:

Increase safety and maintenance scheduling monitoring an early warning indicator.

What it does:

Use data from sensors to predict EGT, and monitore differences between prediction and observation. Behavior through time help to detect potential slow deterioration on the engine, and difference analysis help to detect unusual behavior of the engine. These informations will help to detect severe damage, increasing safety, and schedule maintenance in case of slow deterioration. In a second time, EGT can help to predict fuel consumption, as high EGT is correlated to high fuel consumption.

How we built it:

Step1: Discover and think about data meaning. Provide sense to data and choose the ones which could provide added value in airplane safety and maintenance Step2: Choose useful outputs needed for UI Step3: Design the prediction model (using R and java). Step4: Provides output to front-end Step5: Design of the UI in Seed Application Step6: Display results Step7: Deploy in Predix

Challenges we ran into:

GE aviation challenge

Accomplishments that we're proud of:

Succeed in providing a good prediction model using data from sensors
Succeed in deploying in Predix

What we learned:

Collaboration with front-end developers, coordination between the team members, data analysis, deal with sensor data, prediction and anomaly detection, some software tools, technical background about airplanes engines.

What's next for airplane engine

Two major axis of improvement: Analysis: Explore more in detail variation through times Increase the interpretability of results studying the CHT/EGT couple Display: Improve the UI (Choice of airplane and engine, period, visual warning) Integrate more meaningful indicators thanks to new analysis