Towards Population-Based Structural Health Monitoring, Part VII: EOV Fields – Environmental Mapping

Abstract

In a population-based structural health monitoring setting, data from one structure in a population, where the health state is known, may be used to make inferences about the health state in any nominally-identical structure. Any deviation from the learned ‘healthy response’ potentially indicates damage. However, as in standard applications of structural health monitoring, the healthy response from different structures also varies with the changes in environmental conditions across the population. This paper investigates the modelling of the change in environment across a population of structures located in one geographical region, such as a wind turbine farm. A data-driven mapping method (based on Gaussian process regression) will be introduced that aims to quantify and normalise variation stemming from the environment, such that the remaining response is only sensitive to damage or performance anomalies. The way in which environmental maps are constructed and implemented is demonstrated via a case study from an offshore wind farm. The ideas introduced here will constitute a field in the framework of population-based structural health monitoring presented elsewhere in the conference.