Multi-phase adaptive methodology for mitigating environmental and operational variability in slowly changing time-variant engineering structures

Abstract

This work presents a multi-phase adaptive methodology to mitigate environmental and operational variability (EOV) in slowly changing time-variant engineering structures. By employing stochastic modelling of damage sensitive features (DSF), the approach effectively mitigates EOVs, adapts to structural states, and quantifies uncertainty. A hyperparameter regulates data composition from two proximate, similarly designed structures, capturing EOV-DSF interdependencies across evolving states. The study reveals that long-term monitoring introduces distinct structural phases with similar, but not identical, behaviour, challenging the application of a singular and static model due to the inherent time-variant nature of structure evolution. Applied to an offshore wind turbine undergoing structural integrity evolution, the outcomes underscore the adaptability of the methodology. The results demonstrate that combined data-models can be adjusted to different structural phases over time, yielding more accurate representation compared to traditional methods reliant on static data-models. The multi-phase adaptive methodology proves to be more efficient in delivering less variable and more robust DSFs for the long-term monitoring of slowly changing, time-variant engineering structures.