Optimal sensor placement and assessment for modal identification

A review is presented on a wide range of optimal sensor placement (OSP) strategies and ways of evaluation. An assessment-guided optimal sensor placement algorithm based on minimal root mean square is proposed to give quantitative comparison on these options. The algorithm is applied to an example of a cantilever beam, as well as to a jacket-platform where model uncertainty and measurement error are included simultaneously. The results reveal the influence of the number of desired sensors and the number of modes of interest. It is shown that the optimal sensor placement selected by the proposed strategy yields the best modal orthogonality of the corresponding mode shapes, and competitive results are derived on the other two assessment standards, i.e. the condition number of the reduced mode shape matrix and the determinant of the Fisher information matrix. The conclusions also show that the effective independence method keeps the largest determinant, and there seems no regular trend on the condition number but unexpected biased solutions can be derived when unreasonable combination for number of wanted sensors and modes of interest are chosen. The proposed minimal root mean square algorithm is capable of avoiding this drawback.