A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Developments in sensor techniques have enabled continuous monitoring of the health of operating systems, providing valuable condition data that can enhance failure predictions and inform maintenance decisions. Most condition-based maintenance strategies assume a known type of deterioration process, but in practice, this is generally not known. Therefore, an alternative way is to make maintenance decisions directly based on the available condition and failure data. The main contribution of this thesis is to present fully data-driven approaches for condition-based maintenance optimization in scenarios where limited condition and failure data is available, without assuming a specific type of deterioration process or failure level. The benefits of these approaches are explored, demonstrating that the resulting maintenance policies quickly converge to optimal or near-optimal solutions, despite the challenges posed by only having limited data. Furthermore, these approaches demonstrate robust performance across condition data from various deterioration processes, highlighting the flexibility and reliability of the approaches. The developed fully data-driven approaches provide practical value in modern manufacturing industries, where condition data is continuously collected. Due to their adaptability, the approaches can be effectively implemented in real-world systems. The interpretability of the approaches enables stakeholders to understand and trust the maintenance decisions, fostering their integration into decision-making processes.
Developments in sensor techniques have enabled continuous monitoring of the health of operating systems, providing valuable condition data that can enhance failure predictions and inform maintenance decisions. Most condition-based maintenance strategies assume a known type of deterioration process, but in practice, this is generally not known. Therefore, an alternative way is to make maintenance decisions directly based on the available condition and failure data. The main contribution of this thesis is to present fully data-driven approaches for condition-based maintenance optimization in scenarios where limited condition and failure data is available, without assuming a specific type of deterioration process or failure level. The benefits of these approaches are explored, demonstrating that the resulting maintenance policies quickly converge to optimal or near-optimal solutions, despite the challenges posed by only having limited data. Furthermore, these approaches demonstrate robust performance across condition data from various deterioration processes, highlighting the flexibility and reliability of the approaches. The developed fully data-driven approaches provide practical value in modern manufacturing industries, where condition data is continuously collected. Due to their adaptability, the approaches can be effectively implemented in real-world systems. The interpretability of the approaches enables stakeholders to understand and trust the maintenance decisions, fostering their integration into decision-making processes.
Data-driven condition-based maintenance optimization given limited data
Cai, Yue (author)
2025-01-01
Cai , Y 2025 , ' Data-driven condition-based maintenance optimization given limited data ' , Doctor of Philosophy , University of Groningen , [Groningen] . https://doi.org/10.33612/diss.1220543548
Book
Electronic Resource
English
Pavement Network Maintenance Optimization Considering Multidimensional Condition Data
| Online Contents | 2012
Prediction of Railway Track Condition for Preventive Maintenance by Using a Data-Driven Approach
| DOAJ | 2022