Fault diagnosis of roller bearing feature subset select based on greedy algorithm

Because RST´s ability of data reduction, feature subset selection was translated into the process of data reduction. The condition attributes and decidation attributes of the diagnosis system were reducted, and we received the best training swatch which were cleared up the information of redundance and repetition. Greedy algorithm is a method of discretion and a algorithm of attribute reduction. In the article, fault diagnosis data of roller bearing was discreted and was reducted its attribute. The simple and reliable diagnosis rulers were received, and testing samples ralidated the reliability of the rulers.