Uncertainty-aware Optimal Design of Sensor Configurations for Structural Health Monitoring

The development of digital twins for structural dynamic systems operating under uncertain environmental and operational conditions is important for building reliable data-informed computational models. Data collected from a monitoring system can help build a reliable digital twin that cover the various linear and nonlinear mechanisms activated under different excitations (seismic, wind, wave, etc), monitor the state (healthy or deteriorated) of the structure, as well as reconstruct the response time histories (e.g. drifts, accelerations, strains and stress) at unmeasured locations for evaluating safety-related indices. This work outlines recent developments on the methods for selecting the optimal type, location and number of sensors for building reliable digital twins and monitoring the healthy and damaged state of structural systems (e.g. buildings, bridges, offshore structures, wind turbines, etc), considering the uncertainty in the temporal and spatial variability of the loads. The challenges associated with the design of the sensor configuration to gain information from different competing monitoring tasks are investigated. Such monitoring tasks include modal identification, model selection and parameter estimation, damage identification and virtual sensing. Using information theoretic measures, the optimal design of the sensor configuration is formulated as a multi-objective optimization problem with each objective associated with the information gain from each monitoring task. The optimal design accounts for the modelling, measurement and loading uncertainties that are present at the experimental phase of designing the sensor system to monitor a structure. Methodologies are also presented to trade-off information provided from a monitoring system versus the cost of installing and maintaining the sensor network. Applications on selected structural systems are used to demonstrate the proposed optimal sensor configuration design framework and its effectiveness.