A NASA turbofan simulated engine dataset with 100 engine failures was made available. We demonstrated predicting the impending failure 50 cycles in advance.
A NASA turbofan simulated engine dataset with 100 engine failures was made available. Each of the engines had 24 sensor parameters.
Challenge
All 100 engine data has less than 1% deviation from normal operating data. To the human eye there were no signs or warnings of impending failure. Furthermore, the data was unlabelled and had no historical context.
Solution
Amygda's methods were applied to the engines and built an orchestration of algorithms, from our library of over 200 algorithms.
We found data driven insights and turned them into actionable insights to predict and prevent further engine failures.
Impact
Using Amygda's methods we were able to clearly demonstrate damage and impending failure 50 cycles in advance.
Our platform could be implemented quickly to predict and prevent these failures in the future.
Outcome
100% Detection
Ability to predict and prevent jet engine failures and avoid disruption
50 Cycles early
Allowing the customer to plan repair or replacement in advance
Validation
Our empirical methods without physics based modelling work
Get a competitive edge over your competitors. Let us enable disruption-free industrial operations for your business.
Contact us
Contact us to discuss how we can work together to remove disruption from your industrial operations and reduce cost.