| Summary: | This project focuses on select optimal sensor configuration placement by computational and theoretical matters. A statistical methodology is demonstrated to obtain the best sensor location(s) in a structure with a purpose of picking out the most instructive measured of the parameters which shows the structure behavior. This methodology is also can be used in model updating, structure damages detecting in an earlier stage and response prediction. Information entropy, based on nominal model analysis, generate the results for the selection of optimal locations by indicating certain measured uncertainty in the mode. These uncertainties are generated by Bayesian Methodology, which reduce the measurement of entropy among all available sensor configurations to obtain the best sensor locations. Therefore, this methodology can handle the inevitable uncertainties properly in model parameters and also increase the accuracy of the prediction. There are two types of uncertainties in modelling which will results in the identification of statistical system, namely prediction error and parameter uncertainty. These will be solve by probability models which are constructed through heuristic algorithm. This project based on a twenty-nine degree of freedom (DOF) truss structure (bridge) with large model uncertainties, which is illustrated by determine the optimal sensor configurations using modified information entropy measure and Monte Carlo simulation. The performance of the model updating will be improved when more number of sensors are being used.
|
|---|