Case Study: Airbus detects anomalies in ISS telemetry data with TensorFlow TFX

How Airbus Detects Anomalies in ISS Telemetry Data Using TFX

Airbus needed to automate the detection of anomalies within the vast volume of telemetry data from the International Space Station's Columbus module. Manually monitoring thousands of data streams was an immense challenge for their engineers. To address this, they partnered with TensorFlow and used TFX to build a scalable machine learning solution for anomaly detection.

TensorFlow implemented a solution using a TFX pipeline on Kubeflow to train LSTM autoencoder models on historical data. This system processes real-time telemetry streams to automatically identify anomalies and alert experts, significantly speeding up resolution times. The TensorFlow solution is modular, allowing for the training of separate models for different subsystems, and is designed for a seamless migration to a private cloud, enhancing both security and operational efficiency.

Airbus

Marcel Rummens

Airbus