Overview
Click on the menu bar items to navigate to chapters
Click here for PDF version
Progressive Visual Segmentation of
Vibration Signals to Detect Change Points
Julian Rakuschek, Johanna Schmidt, Udo Schlegel, Michèle Posch, Jutta Isopp, Tobias Schreck
VMV 2026 - 25.09.2026
Engine
Sensor
Vibration signal
Find long-term deviations
Find long-term deviations
Find anomalies (short-term, unexpected)
Find anomalies (short-term, unexpected)
Predict faults
Predict faults
Condition Monitoring
Predictive Maintenance
Anomaly Detection
The transitions between segments mark change points
Ali, Mohammed et al. “TimeCluster: dimension reduction applied to temporal data for visual analytics.”
The Visual Computer 35 (2019): 1013 - 1026.
Bernard, J. et al. “TimeSeriesPaths : Projection-Based Explorative Analysis of Multivariate Time Series Data.”
International Conference in Central Europe on Computer Graphics and Visualization (2012).
Schmidl, Sebastian et al. “DendroTime: Progressive Hierarchical Clustering for Variable-Length Time Series.” International Conference on Extending Database Technology (2026).
Example: 30 kHz (30 000 samples per second)
Assumption: 8 byte per sample
Manual scanning is tedious!
We iteratively construct an overview
In many cases a small fraction of the data suffices!
No!
We only save a FFT histogram per sample into the database.
Time Series Projections
We obtain a segmentation
5 samples
10 samples
25 samples
50 samples
5 samples
10 samples
25 samples
50 samples
(Accelerated video footage)
The result: Sharp boundaries
The result is a uniform distribution of samples
Pick samples in chronological order
(Accelerated video footage)
With two domain experts from the field of vibration analysis
Round 1 / 3
Round 2 / 3
Round 3 / 3
Early analysis results:
Final analysis results:
Julian
Johanna
Udo
Michèle
Jutta
Tobias
Open Source
PRESENT and ENVIRON-HYDRO
Slides