intelligence driven innovation

The APU is not only a great power source to generate onboard electricity and heat, but it is also the lifeline of the aircraft when the main engines stop working. Therefore airlines rely fully on a perfect working APU. In the case of the APU malfunctions, it has to be repaired immediately. This harms the continuity of the operation.

As one of the leading organizations in aircraft component repair, EPCOR, is fully aware of the impact of a malfunctioning APU. They, therefore, offer to their clients some years a predictive maintenance tool that enables airlines to monitor the status of their fleet.

We helped EPCOR building a deep learning model capable of predicting APU malfunctions of the Boeing 737. For more information watch this video.

  • idea

    Predict Auxiliary Power Unit (APU) failures 200 cycles in advance for a renowned aircraft maintenance company

  • data

    2 years of time series data from 20 different APU sensors

  • technology

    Deep learning sequential model (LSTM cells)

  • scope

    Modelling, testing, and integration

  • planning

    Realized in 8 weeks



Increase of the clients detection rate compared to the traditional machine learning model they used before.


Decrease in aircraft engine maintenance costs