SVM-Association Rules Extraction Applied inMetallurgical Industries

摘要

A SVM- Association Rule Extraction (SARE)rnalgorithm, which has been studied by Multi-classrnSVM and SVDD methods, is proposed to solve arnpractical optimization problem of process industry inrnthis paper. SARE algorithm is constructed on thernbase of prototypes and support vectors (SVs) underrnsome heuristic limitations. The proposed algorithm isrnapplied to a Wet gas Sulphuric Acid (WSA) processrnand the relationships between the key processrnattribute and objective attribute are obtained byrnSARE. Using existing domain knowledge aboutrnWSA process, most of the obtained association rulesrncan be explained and two of the rules show importantrninformation.