| Autor: |
Simon Leohold, Hendrik Engbers, Michael Freitag |
| Rok vydání: |
2020 |
| Předmět: |
|
| Zdroj: |
Dynamics in Logistics ISBN: 9783030447823 |
| DOI: |
10.1007/978-3-030-44783-0_23 |
| Popis: |
An essential requirement for climate-friendly and sustainable transport and logistics services is the cost improvement of rail freight services. Maintenance in rail freight is a major cost driver. Therefore, the goal is to reduce costs by digitalization and state-of-the-art maintenance approaches. In this paper, we present an approach for an individual predictive maintenance system for diesel engines of rail vehicles. It is a data-driven approach that leverages data characteristics from historical and current time series data from the Engine Control Unit (ECU). The proposed methodology applies a meta-learning technique to select the most suitable forecasting model for each engine component in order to determine the time of failure. The meta-learning technique allows the methodology to be applied to other engine series. As a result main fault classes of an engine type have been identified and the corresponding potential based on the analysis of historically corrective and preventive measures are presented. Further analysis shows that the lifetime of turbochargers and the injection system are insufficiently exploited. |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
