Optimal sensor placement for response prediction of civil engineering structures subjected to future uncertain dynamic loadings

Computational and theoretical issues in selection of optimal sensor configuration are addressed in this project. A statistical methodology is being put forward to optimally locate best sensor configuration within a structure with the objective of selecting the most comprehensive measured figures of the parameters. The optimal configuration selection depends on information entropy which is based on nominal model analysis that indicates specific measured uncertainty in model parameters. Hence, information entropy is valued as an evaluation indicator and the Bayesian statistical methodology is adopted to compute the parameters’ uncertainties. The entropy measure is reduced for potential sensor configurations, applying through a proposed efficient and accurate heuristic algorithm to generate an optimal sensor configuration. Moreover, this methodology is applicable to model updating, response prediction and early detection of structural damages. Two general types of modelling uncertainties, parameter uncertainty and prediction error, arise in the statistical system identification and need to be computed by probability models constructed through heuristic algorithm. Large model uncertainties cases are being addressed in this project with optimal sensor configurations being determined by modified information entropy measure and Monte Carlo simulation. Measured assessment as well as correlation of model parameters’ values are obtainable through the different sensor placements in each configuration. The project is demonstrated with a nine degree of freedom (DOF) building model.Bachelor of Engineering (Civil