Colloquia

Samenvatting

The implementation of predictive maintenance (PdM) in complex technical systems is often hindered by uncertainty regarding component suitability, data adequacy, and the feasibility of different PdM tasks and modelling approaches. While various PdM models and approaches exist, organizations—particularly in safety-critical and data-constrained domains such as naval applications—lack structured guidance for determining whether PdM is feasible. In addition, there is limited guidance on how to determine PdM feasibility for specific components, PdM tasks, and modelling approaches, and how to select and use these approaches effectively. This thesis addresses this gap by developing and evaluating a generic, operational decision-support tool for predictive maintenance feasibility assessment.

Using a design science research methodology, a three-module decision-support tool was designed to give guidance on the PdM decision-making and the feasibility assessment process. The first module guides critical component selection and performs an initial PdM suitability and feasibility screening based on component criticality, sensor availability, and sensor-failure mode observability. The second module refines this selection by prioritizing components with relevant failure history using historical maintenance data. The third module provides a structured, quick-scan feasibility assessment for different PdM tasks—fault detection, fault diagnosis, health assessment, and prognosis—across rule- or knowledge-based, physics-based, and data-driven modelling approaches, using explicit data, model, and knowledge requirements.