Application of FCM clustering based rough sets on steel rolling process

This paper presents a model for predicting steel mechanical property based on rough sets and fuzzy c means clustering. Rough sets is an intelligent method, which can only be applied to data table with discrete attributes. However the practical data set is normally continuous, and rough sets cannot be used directly. FCM clustering is used to transform the continuous attributes to discretized ones and a discretized decision table can be got. Rough sets reduce the discretized decision table to discover significant attributes of a data set and filter out those attributes which are unimportant. Finally, to verify the validity of the proposed method, it is used for practical data acquired from some steel works, and the simulation results show that the attribute reduction contains the same information as the original one.