A Hybrid Bayesian Compressive Sampling and Radial Basis Function Algorithm for 2D Spatial Reconstruction of Concrete Strength Using Sparse Measurements

Abstract Obtaining the spatial distribution of concrete properties for large-scale concrete structures through in situ tests is challenging due to the difficulty of collecting an adequate number of representative samples and the need for strategically selecting optimal sampling locations. This paper presents an advanced computational algorithm for reconstructing the 2D spatial distribution of concrete properties using sparse field measurements. The developed algorithm, named “advanced Bayesian compressive sampling” (ABCS), integrates Bayesian compressive sampling (BCS) with the radial basis function to overcome the limitation of BCS under systematic grid-based sampling method. ABCS ensures that the reconstructed field not only captures the underlying stochastic variability but also passes precisely through measured sampling points. A generalized design aid was developed to complement the algorithm by quantifying the prediction accuracy as a function of sampling density and concrete material variability using extensive random field simulations. The developed algorithm and design aid were validated on two real-life case studies of degraded concrete structures. The results demonstrated that the method can accurately reconstruct the 2D spatial distribution of concrete compressive strength with an acceptable margin of error while significantly reducing the required sampling points. Moreover, the calculated errors were within the range of what was presented in the design aid, confirming both the practical and analytical reliability of the proposed method.