Welding type oriented association rules acquisition based on rough sets

Large quantities of data are accumulated with the wide applications of computer aided process planning software in Body In White(BIW). To acquire the potential and valuable process knowledge from these data, a methodology of welding type acquirements based on the rough set theory and association rule technique are proposed. The attributes related to the welding type of parts are analyzed, among them quantitative attributes are discretized, and a decision table for the selection of welding type is generated. Rough set theory is employed to remove redundant attributes. Apriori algorithm is used to extract frequent attribute item sets. In order to reduce generations of the redundant candidate itemset, two items which belong to the different attributes are joined. The strong rules whose consequents are welding type are generated according to the minimal confidence. Welding data of a BIW are processed. Generated association rules have great reference value to the selection of the welding type.