Artificial intelligence methods (AI methods) have already proven their high potential in many applications. The aim of the research project "KI-MeZIS" (Ger. abbr., AI methods in structural health monitoring and condition-based maintenance of rail vehicle structures) is to tap this potential for the monitoring and evaluation of rail vehicle structures and thus contribute to improving the performance of rail transport. The joint project is funded by the German Federal Ministry for Economic Affairs and Climate Action and the involved project partners are Deutsche Bahn AG and DB Netz AG within the sector initiative Digitale Schiene Deutschland, the Institute of Vehicle Concepts of the German Aerospace Center e.V., Industrial Analytics GmbH and the Institute of Machine Components of the University of Stuttgart. The project's investigations are based on data from sensors installed on the outer panelling and load-bearing elements of Deutsche Bahn AG's research and test vehicle advanced TrainLab (aTL, type ICE TD 605) and their evaluation using AI methods. The conducted research encompasses the use of the sensors and their data for collision detection, structural health monitoring, condition-based maintenance and needs-oriented design of future rail vehicles. Within the scope of the project the Institute for Machine Components’ focus is on the development a sensor-based system for structural health monitoring and needs-based maintenance for the metallic load-bearing structure in the front area of the test vehicle which can be subject to crash and impact loads. The system is based on sensors for structural health monitoring and a database containing information on the structural condition of the monitored components in the event of a collision and recommendations for necessary maintenance measures. Its purpose is to provide the utilised AI methods for data-scanning, machine learning and decision-making processes with (training) data. Therefore, it enables them to identify anomalies in the sensor data in the event of a collision and to link the anomalies with the associated damage and necessary maintenance measures. At NAFEMS World Congress 2023 the developed sensor system for structural health monitoring and the pursued methodology for the creation of the database’s content regarding structural information are presented. The developed on-board sensor system for structural health monitoring consists of strain gauges and a suitable data acquisition unit. To limit the scope of the project’s investigations, representative components of the metallic load-bearing structure in the area of the test vehicle’s front were selected, such as the train’s pilot and the coupling’s connection to the car body. The sensor system monitors the equipped components during the operation of the vehicle and generates machine learning data during test rides representing sensor signals in absence of collision events. The applied sensor types of the system and their positioning were chosen according to the experience gained from previous projects, the analysis of the structural behaviour of the monitored components using FEM and project-related constraints. In order to generate the collision-related contents of the database, the structural behaviour of the monitored components in the event of a collision is investigated by means of extensive crash and impact simulations based on FEM. For this purpose, a variety of representative load cases is defined with the aim of exposing the monitored components to realistic collision scenarios and loading conditions. The defined load cases are therefore derived from real-world crash and impact events which are investigated in this project as well. The effects of these load cases on the monitored components and their resulting structural state are analysed regarding occurring stresses, strains, damage, etc. and according information is recorded in the database. To validate the simulation results and database contents, bench tests will be carried out for which original components of the test vehicle are available.