Failure analysis and diagnostics for railway trackside equipment

In the future, the dependability of the railway system must be improved to accommodate higher train speeds, greater axle loads, and increased service frequency. One way of increasing the dependability is through improved maintenance and failure management. The collection and processing of information to ascertain the condition of railway infrastructure allows remedial action to be undertaken in a more efficient manner. This paper describes the development of a condition monitoring system based on pattern recognition for detecting and diagnosing point machine failures. It is shown that initially, when data is acquired from a point machine, it is difficult to detect faults. However, upon the application of a moving average filter, fault conditions can be more straightforwardly detected and diagnosed. Using data collected from a set of points under test it is shown that all potentially tested failure modes can be successfully detected and diagnosed.