An algorithmic approach for maintenance management based on advanced state space systems and harmonic regressions

Point mechanisms are special track elements which failures results in delays and increased operating costs. In some cases such failures cause fatalities. A new robust algorithm for fault detection of point mechanisms is developed. It detects faults by comparing what can be considered the ‘normal’ or ‘expected’ shape of some signal with respect to the actual shape observed as new data become available. The expected shape is computed as a forecast of a combination of models. The proposed system deals with complicated features of the data in the case study, the main ones being the irregular sampling interval of the data and the time varying nature of the periodic behaviour. The system models are set up in a continuous–time framework and the system has been tested on a large dataset taken from a point mechanism operating on a commercial line.