Clustering and association rules in analyzing the efficiency of maintenance system of an urban bus network

摘要

Maintenance has always been considered as an important part of both manufacturing and service systems and yet a costly practice. The purpose of this study is to analyze the efficiency of the maintenance activities in a maintenance system comprising of independent components, using the collected data in process. For this purpose, a three-stage method was followed. First, at the initial data preprocessing stage, after the data purification, new operating fields were defined. The data was integrated in a final matrix which was used as an input for the modeling phase. At this stage, using one of the clustering algorithms i.e. k-means, the maintenance data was clustered so that homogenous clusters of the components i.e. buses, were formed. Then using the Euclidean distance, the distances of the clusters from the ideal status were found and clusters were categorized and named accordingly. In the last part of the modeling stage, while having the clusters as target, Apriori algorithm was used to identify the rules (conditions) which had caused each record to be placed in each specific cluster and thereby to find a way to assess the efficiency of the maintenance system and activities. At the 3rd stage and on the basis of the extracted rules, necessary steps were proposed to eliminate the conditions which lead records to be placed in the clusters comprising records of bad conditions. The method is explained in a case study of the maintenance system of an urban transportation bus network.