Skip to content

Intelligent maintenance solution provides more reliable railroad trafic

Published: 22 February 2012

By using methods of intelligent fault diagnosis for analysis of maintenance data in the rail system, such as electric motors in the switching systems, train services can be more secure, according to new research at Luleå University of Technology. Well known is the fact that the Swedish rail often have major problems in cold winters, a problem that can be reduced.

- Accurate and efficient fault diagnosis can ensure that the one in charge of managing, and the operator of railroad trafic takes the right maintenance decisions, and by that enhancing rail system availability and reduce maintenance costs, says Yuan Fuqing, PhD at Luleå University of Technology.

In a comparative study between two methods of analysis, Yuan Fuqing, a doctoral student in the Department of Operation, maintenance and Acoustics at Luleå University of Technology have found an optimal way to early identification of defects in the rail system. He has compared the Support Vector Machine (SVM) with other well known techniques for error analysis in the operation and maintenance area, such as Artificial Neural Network (ANN). By various ways to improve SVM Yuan Fuqing has been able to determine that it is the best method.

- SVM is an excellent technique to identify and recognize different patterns, but to get a proper diagnosis from the technology, it is necessary to extract appropriate features from it which I have done in various ways to optimize the method, he says.

The advantage of SVM is that one can study complex data at a detailed level, where they can not be interpreted, but also transform into a higher dimension where they are part of a whole and can be separated out and be understood. This in turn provides an opportunity to at an early stage, and cost-effective, be able to make an errordiagnosis that allows one to identify errors that are about to occur.