Clustering of Steel Strip Sectional Profiles Based on Robust Adaptive Fuzzy Clustering Algorithm

keywords: Fuzzy clustering, robust, adaptive operators, outlier mining, steel strip profile
In this paper, the intelligent techniques are applied to enhance the quality control precision in the steel strip cold rolling production. Firstly a new control scheme is proposed, establishing the classifier of the steel strip cross-sectional profiles is the core of the system. The fuzzy clustering algorithm is used to establish the classifier. Secondly, a novel fuzzy clustering algorithm is proposed and used in the real application. The results, under the comparisons with the results obtained by the conventional fuzzy clustering algorithm, show the new algorithm is robust and efficient and it can not only get better clustering prototypes, which are used as the classifier, but also easily and effectively detect the outliers; it does great help in improving the performances of the new system. Finally, it is pointed out that the new algorithm's efficiency is mainly due to the introduction of a set of adaptive operators which allow for treating the different influences of data objects on the clustering operations; and in nature, the new fuzzy algorithm is the generalized version of the existing fuzzy clustering algorithm.
mathematics subject classification 2000: 68Txx, 68Uxx
reference: Vol. 30, 2011, No. 2, pp. 357–380