Application of FP_Growth Algorithm of Sequential Pattern Mining on Container Maintenance Components Association

Using the available historical maintenance data to reveal the patterns and laws behind the frequent damages would provide a possible solution to the maintenance issues. This study collected 9 years container maintenances record data from China Railway Container Transport Corp, Ltd (CRCT). Based on those data, establishes a maintenance component chain for the maintenance parts of the maintenance operation, and converts it into a sparse matrix format data set according to the algorithm requirements. FP_growth (Frequent Pattern) algorithm is used to analyze the data set. According to the actual situation, the value of the confidence and lifting degree are set, and the 178 strong association rules are obtained. The results indicate that the components of the doors are most frequently damaged parts and there are strong relationships among the damages of components of rear end and doors.