Parameter Estimation of Structures from Static Strain Measurements. II: Error Sensitivity Analysis

Abstract

A method for the parameter estimation of structures was developed in a companion paper. It used subsets of static applied forces and strain measurements and successfully identified structural element stiffnesses. In the presence of noisy measurements, the selection of these subsets drastically affect the accuracy of the parameter estimates. In addition, there are a large number of combinations of applied force and strain measurements to be considered. Therefore, a heuristic method is presented for choosing a small subset of forces and strain measurements so as to have the least sensitivity to measurement noise. Once chosen, using a Monte Carlo analysis the subset of measurements is analyzed to determine an input-output error relationship. This heuristic method can be used to design near-optimal experiments for the nondestructive testing and parameter estimation of structures. A truss and a frame example are presented and a small and noise-tolerant subset of forces and strains is successfully selected for possible nondestructive testing.