Hierarchical Clustering applied to Measured Data Relative to Pantograph-Catenary Systems as a Predictive Maintenance Tool

A hierarchical clustering algorithm, based on Fuzzy C Means, is developed and applied to a set of measured data (voltages and currents) collected by the pantograph of high speed trains in order to detect the presence of electric arcs and classify their magnitude. During the test runs the electric arc has been recorded by a photosensitive device, and is used in this analysis to evaluate the effectiveness of the clustering procedure. The results show that the so obtained clusters are effectively related to the presence and magnitude of electric arcs and that the technique could be used as a tool for preventive maintenance.