Supervising MultiCut aggregates by special neural nets

In this paper we present a tool, supervising a multi-cut aggregate working in the German factory for pipeline production Salzgitter Mannesmann Line Pipe GmbH by CI-based methods. To achieve this aim we use special pre-processing strategy of the DLS-Spectra and topological closed SOMs used as “unsharp classifiers in a defined range” by using the principle of “Computing with Activities (CWA)”. We will show that this strategy leads to an early detection of machinery faults, creeping trends of wear effects and sudden uprising of so far unknown sensor states.