An innovative sensor network consists of soft elastomeric capacitors (SECs) has been proposed for large scale civil infrastructures such as wind turbines. Each SEC unit is able to convert a change in local strain into a change in capacitance due to geometric deformation of the patch. Since each SEC unit measures unidirectional strain under the covered surface, a method is needed to be developed in order to decompose strain into principle components.

To perform the strain decomposition task, a strain surface fitting algorithm using polynomial interpolation functions has been developed. Firstly, the proposed algorithm was validated on both symmetrical and nonsymmetrical plates under various loading conditions. The strain polynomial functions used for plates were assumed based on the classic plate theory. By minimizing the errors between real and estimated strain values, a coefficient matrix was estimated and used to estimate bi-dimensional strains. The results of predicted strains showed good agreement with real strain data extracted from the finite element models. Once the algorithm was validated, multiple ways of sensor arrangements were performed in order to investigate the effects of sensor placement on estimating accuracy. It is found that the estimating accuracy can be improved if either the inner sensors were staggered or the amount of boundary sensors was reduced.