During the ride in a light rail vehicle, passengers perceive permanent acceleration exposure. In contrast to the main-line-track the acceleration’s magnitude, frequency and characteristic arise from the highly non-stationary driving operation. Especially trams move frequently into curves without transition curves and the curve radii are often small. This leads to big peaks in both acceleration and jerk values. Besides these effects, the passengers experience extensive fluctuations in the longitudinal dynamics due to stop and go events at traffic lights and stations. This forms a subjective ride comfort impression of each passenger. The evaluation of the ride comfort in rail vehicles can be carried out using numerous methodologies, that have been developed within the scope of the main-line-track. Thus, particular incompatibilities arise when these methods should be applied to BOStrab-vehicles that lead to the motivation of this work. A ride comfort evaluation method will be developed that is specifically tailored for fitting the instationary conditions of the BOStrab-operation. It can be used to predict the ride comfort perception of seated and standing passengers. With the help of machine learning two different algorithms are developed. The resulting models can predict the mean ride comfort perception of the passengers based on different features that are derived from three-dimensional accelerations, measured on the floor assembly of the vehicle. All subjective passenger ratings are collected using a self developed smartphone application that is used in several measurement campaigns. A final examination of the model’s exactness based on validation measurement data shows a very good performance and precision. In order to embed the algorithms into an overall evaluation methodology the dynamics and traffic mode of the BOStrab-operation was analyzed statistically. To formulate adequate boundary conditions for the overall ride comfort evaluation methodology, additional information about the operation mode based on further measurement data is used. The new developed ride comfort evaluation method for BOStrab-vehicles closes several gaps of the state of the art and technology in the field of ride comfort evaluation of light rail vehicles.
Schwingkomfortbewertung in Straßenbahnen mittels Maschinellen Lernens; 1. Auflage
Ride comfort evaluation in trams using machine learning methods
2023-01-01
Online-Ressource : Illustrationen, Diagramme pages
Dissertation, RWTH Aachen University, 2022; Aachen : Apprimus Verlag, Schriftenreihe Schienenfahrzeuge und Transportsysteme 5, Online-Ressource : Illustrationen, Diagramme (2023). = Dissertation, RWTH Aachen University, 2022
Miscellaneous
Electronic Resource
German
Semantische Objekterkennung mit Methoden des Maschinellen Lernens
Tema Archive | 2013
|European Patent Office | 2025
|