Monitoraggio di catenarie ferroviarie mediante analisi statistica di segnali di misura. = Monitoring of railway catenaries through statistical analysis of measurement signals.

In the last years, railway networks have had considerable expansion on all continents: both for transport of passenger and for commercial purposes. Therefore, concerning future role that rail transport will have it is necessary to study ways that allow reliable working and reduce maintenance costs. One of the fundamental aspects in achieving these targets is the study of the interaction between pantograph and catenary, crucial for smooth power transmission and avoiding wear. . In particular , thanks to the development of artificial intelligence technologies and increasingly precise image-taking systems, several fault detection and diagnosis techniques for high-speed train are used nowadays; these technologies allow to identify possible anomalies in operating conditions, compared to the nominal operation regime. The aim of this work is to study different statistical parameters, obtained from the measurements of the pantograph displacement or the accelerations of the supports, and to track their evolution over time in order to identify a possible anomaly. In the following, by means of non-linear finite element model, which implements the presence of one pantograph in order to study its interaction with the overhead contact line, possible anomalies -such as decrease in tension, stiffening of a support, breakage of a droppers or decrease of wire contact cross section - were simulated one at a time. The study on the parameters was conducted through the development of a graphical interface, which displays the zones of the catenary for which the parameters highlight any possible anomalies. An initial step considered noise-free simulations with one anomaly at a time., Subsequently, various noise levels were introduced, followed by studying the behavior of the parameters in presence of several concurrent anomalies. For future studies, displacement or acceleration signals taken from real measurements could be analyzed. Furthermore, the methodology could be extended to the presence of two pantographs, which changes the dynamic of the interaction and influences the statistical parameters. Concerning the estimation of contact force, would be interesting to develop an appropriate contact model for the pantograph that uses as input the measured displacement of the pantograph head.