Case Study: University of Stuttgart achieves AI-based time series anomaly detection with MathWorks

AI-Based Time Series Anomaly Detection for Cyber-Physical Systems at the University of Stuttgart

The University of Stuttgart needed a more general way to detect anomalies in complex cyber-physical systems, where faults in sensors, computing hardware, or networks can quickly lead to failures and safety issues. To address this challenge, the team used MathWorks tools, including MATLAB, Simulink, Deep Learning Toolbox, and Statistics and Machine Learning Toolbox, to build a time series anomaly detection platform called TSAD.

MathWorks helped the University of Stuttgart create a workflow that imports and preprocesses data, trains and compares statistical, machine learning, and deep learning models, and then deploys the best detector back into Simulink for testing. In a robot case study, TSAD selected a fully connected autoencoder that detected a sensor fault and triggered an emergency shutdown before the robot could drop a tube, showing how MathWorks’ solution improved fault handling and safety, though the case study did not provide numeric performance metrics.

University of Stuttgart’s

Andrey Morozov

University of Stuttgart’s