Diagnosis and prognosis for maintenance management in rail transport

In a research project funded by the DZSF1, diagnosis-based and predictive approaches to cost-efficient maintenance are being developed. 

Efficient maintenance of infrastructure and rolling stock is the basis for safe railway operations and helps to improve the quality and availability of rail transport. The maintenance management of railway companies is gradually moving from damage-based (corrective) and preventive maintenance to condition-based and predictive maintenance strategies. Digitalisation and automation are reinforcing this trend through the further development and spread of technologies used to collect, transmit and process data.

Diagnosis and prognosis are the essential building blocks of predictive maintenance. This is where the research project initiated by the DZSF comes in, which is being implemented by IABG together with the partners Institut für Bahntechnik (IFB) and Havelländische Eisenbahn AG (HVLE).

The first step towards predictive maintenance is to know as much as possible about the current state of a system. Maintenance that has so far been carried out in a corrective or preventive manner can already be optimised if, for example, initial characteristics of developing damage can be detected with the help of sensor technology, thereby enabling a targeted maintenance activity derived from the detected state of deterioration.

Maintenance can be significantly improved even further if the asset's usage history is tracked through permanent and sufficiently closely timed monitoring. If the latter can additionally be assigned to a service life consumption via analytical, model-based or data-based approaches, the remaining useful life can also be estimated with the help of an assumption on future use based on a typical usage spectrum of the asset. This makes it possible to predict the deterioration and to identify the maintenance activities that will likely be required in the future. The considerable expenses and effects of failures due to emergency management, replacement measures, ad hoc repairs and restoration of regular operation can be avoided, as can increased expenses due to preventive, i.e. ultimately premature, maintenance. Individual maintenance activities can thus be coordinated and carried out cost-efficiently in the context of the entire asset management. At the same time, availability increases, which is an essential factor for life cycle costs.

Against this background, concepts are being developed in this project through which condition-based or predictive approaches can be realised and used beneficially for rail transport in the future. The project is divided into several sections::

  • Determination of the state of the art of diagnostic and prognostic methods within as well as outside of rail transport (e.g. in aviation or energy technology) with the classification into physical/model-based, data-based and hybrid models.
  • Evaluation of potentials and limitations of diagnostic and prognostic applications, especially when using model combination and data fusion techniques.
  • Consideration of two exemplary use cases, one for rail vehicles and one for rail infrastructure, in order to emphasise the benefits of the chosen technological maintenance management approach.
  • Identification of further research and development needs in order to be able to foster predictive maintenance concepts in rail transport.
  • Development of recommendations for action to facilitate the migration towards condition-based/predictive maintenance procedures.

The project started on 1 March 2023 and is scheduled to run for three years. The results are planned to be published on the DZSF project website after project completion.


1DZSF: Deutsches Zentrum für Schienenverkehrsforschung