Monitoring Large-Scale Rail Transit Systems Based on an Analytic Hierarchy Process/Gradient-Based Cuckoo Search Algorithm (GBCS) Scheme

Abstract

AbstractCondition monitoring is used as a tool for maintenance management and function as input to decision support. Thus the key parameters in preventing severe damage to railway assets can be determined by automatic real-time monitoring. The technique of radio-frequency identification (RFID) is increasingly applied for the automatic real-time monitoring and control of railway assets, which employs radio waves without the use of physical contact. In this work, a 243-km2 area of Kuala Lumpur was selected. Because of its large size, determining the locations in which to install the RFID readers for monitoring the bogie components in the Kuala Lumpur railway system is a very complex task. The task involved three challenges: first, finding an optimal evolutionary method for railway network planning in order to deploy the RFID system in a large-area; second, identifying the large area that involved functional features; third, determining which station or stations should be given priority in applying the RFID system to achieve the most effective monitoring of the trains. The first challenge was solved by using a gradient-base cuckoo search algorithm for RFID system deployment. The second challenge was solved by determining all necessary information using geographic information system (GIS) resources. Because of the huge volume of data collected from GIS, it was found that the best method for eliminating data was to develop a new clustering model to separate the useful from the unuseful data and to identify the most suitable stations. Finally, the data set was reduced by developing a specific filter, and the information collected was tested by an analytic hierarchy process as a technique to determine the best stations for system monitoring and control. The results showed the success of the proposed method in solving the significant challenge of large-scale area conditions correlated with multi-objective RFID functions. The method provides high reliability in working with complex and dynamic data.