Performance of Sensor Placement Strategies Used in System Identification Based on Modal Expansion

Modal expansion techniques are typically used to expand the experimental modal displacements at the sensor positions to other unmeasured degrees of freedom. Since in most cases the sensors can be attached only at limited locations in a structure, an expansion is essential to assess the condition of substructures such as tanks and pipelines which are attached to this main structure. Most of the optimal sensor placement algorithms for system identification aims to reduce the correlation between the different modes thereby clearly distinguishing between closely spaced modes. In this study, the optimal sensor configuration provided by one such algorithm is studied in the context of modal expansion under the presence of modelling errors and measurement noise. A statistical sensitivity analysis is carried out to measure the correlation between the expanded mode shapes obtained from the optimal sensor positions and those corresponding to the real structure. A cantilever beam and an industrial tower are used as case studies. It is found that this correlation decreases with an increase in both the modelling errors and measurement noise. This forms the basis for a more extensive study aimed at identifying sensor locations which are more robust to modelling errors and measurement noises and thus can be used for a more reliable modal expansion.