On-Line Fault and Anomaly Detection

  • Edwin Lughofer
Part of the Studies in Fuzziness and Soft Computing book series (STUDFUZZ, volume 266)


This chapter deals with on-line quality control systems where measurements and (audio) signals are processed and examined whether they show untypical occurrences, significantly deviation from the normal operation process. The first part of this chapter (Section 8.1) is a natural successor of the previous one, as it deals with the application of on-line identified models at multi-channel measurement systems for early detection of failures in industrial systems. Therefore, a novel on-line fault detection strategy coupled with the usage of evolving fuzzy systems will be presented and its performance demonstrated based on on-line measurements from engine test benches. Section 8.2 deals with the detection of any types of untypical occurrences, also called anomalies (not necessarily faults, but also transient phases), in time series data. Univariate adaptive modelling strategies are applied whose (univariate) responses are used by an integration framework in order to obtain a final statement about the current state and behavior of the production process. Section 8.3 concludes with an application from the signal processing area, dealing with audio signals coming from analogue tapes which should be digitized onto hard disc. There, the detection of a broad band of noises is a central aspect in order to estimate the quality of a tape and to communicate to an operator whether a feasible digitization is possible or not.


Fault Detection Time Series Data Fuzzy Model Anomaly Detection Residual Signal 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Springer-Verlag Berlin Heidelberg 2011

Authors and Affiliations

  • Edwin Lughofer

    There are no affiliations available

    Personalised recommendations