Long title: Capture, monitoring, and evaluation of the actual usage of chain-driven vehicles for optimizing life cycle costs.

Assignment to our (IMW) research priorities:

  • Analysis and Semantic Modeling of Heterogeneous Data Sources
  • Data-Driven Optimization of Maintenance Processes
  • Unstructured Data Analysis and Text Mining in Maintenance
  • Knowledge-Based Maintenance

 

Sponsor: FFG

Duration: 01.10.2023 – 01.10.2026

Abstract:

The usage profile of tracked vehicles varies depending on various factors such as type of use, duration and operational areas. The reciprocal effects of these factors can lead to a rapid increase in maintenance and repair costs, fuelled by numerous sensitive and wear-prone components in track-driven vehicles. The aim of "BEHAVE" is a realistic calculation and optimisation of life cycle costs for tracked vehicles. The objective is to reduce life cycle costs by up to 20% by 2030. For this purpose, heterogeneous data sources are collected as well as transparent and comprehensible usage profiles for tracked vehicles are developed by applying AI-based methods. The automated recording and monitoring of the actual usage and the associated wear and tear of selected components of tracked vehicles allows the real life cycle costs to be identified, predicted and optimised. The result is a proof-of-concept and corresponding demonstrators for predicting the expected life cycle costs based on the real usage profiles of chain-driven vehicles. It is based on an algorithmic model that enables the classification of the usage and stress profile on the foundation of the vehicle data. The individual usage profiles of each vehicle then enables the development of (i) a data-driven decision support system for predictive maintenance planning and (ii) a scheduling model that determines the material consumption depending on the expected vehicle usage.

Results:

Conceptualized comprehensive guidelines for RSF implementation; Formulated optimal sensor placement strategies for enhanced data collection; report on unsupervised algorithms improving RSF modeling; Explored algorithms optimizing RSF combustion understanding; Identified low-dimensional manifolds for efficient RSF reduced-order models; Developed hybrid physics-aware data-driven models and applied multi-fidelity data fusion for RSF systems.

Partners:

Contact details:

Dipl.-Ing. Andreas Steiner

Email: andreas.steiner@tuwien.ac.at

 

 

Image by Freepik