Pattern Matching for Time Series Data

Task

Automatic detection of anomalies in production processes at Daimler to support predictive maintenance. Analysis of thousands of time-dependent signals per production cycle to detect deviations and impending failures at an early stage.

Solution approach

Use of MATLAB for analysis and pattern recognition in time series:

Pattern matching between reference and test signals
Automatic time axis adjustment (compensation for delays)
Detection of significant pattern deviations using correlation and distance analyses

Result

The process reliably detects critical process deviations, enables predictive maintenance and reduces downtime and costs. This makes an important contribution to digital transformation and quality assurance in manufacturing.

You can find our presentation on this topic at the Automotive Conference in Stuttgart here.