Condition Monitoring at Rolling Mills with Data-Driven Residual-Based Fault Detection

Description

In this paper, we propose a residual-based approach for fault detection at rolling mills, which is based on data-driven soft computing techniques. The basic idea is to transform the original measurement signals into a feature space by identifying multi-dimensional relationships contained in the system, representing the nominal fault-free case and analyzing residuals with incremental/decremental statistical techniques. The identification of the models and the fault detection are conducted in completely unsupervised manner, solely based on the on-line recorded data streams. Thus, neither annotated samples nor fault patterns/models, which are often very time-intensive and costly to obtain, need to be available a priori. As model architectures, we used pure linear models, a new genetic variant of Box-Cox models (termed as Genetic Box-Cox) reecting weak non-linearities and Takagi-Sugeno fuzzy models being able to express more complex non-linearities, which are trained with an extended version of SparseFIS. Our approach will be compared with a renowned state-of-the-art approach using PCA components directions based on three different typical scenarios on rolling mill production.

Period	20 Jun 2013
Event title	IFAC Conference on Manufacturing Modeling, Management and Control (MIM) 2013
Event type	Conference
Location	Russian FederationShow on map