Virtual sensors for SHM using isogeometric piezoelectric finite elements

A design of sensor networks for structural health monitoring (SHM) with guided waves poses a hard challenge. Therefore different approaches are possible. A known one is the usage of probability of detection (POD) criteria. Here, areas of potential impact sensitivity are calculated for every sensor which leads to a POD. The number of sensors is increased until a demanded POD is reached. However, these calculations are usually based on finite element methods and underlie different assumptions and approximations which can cause different inaccuracies. These limitations are avoided by using an experimental data basis for virtual sensors in this paper. The paper aims to discuss these issues.

An air-coupled ultrasound scanning technique is used for guided wave investigations. Recorded displacements of a structure surface are used as stimulation of virtual sensors which can be designed by software and positioned within available data field. For the calculation of sensor signals an isogeometric finite element model is used. The virtually bonded layer of the virtual piezoceramic sensor interpolates with non-uniform rational B-Splines (NURBS) the measured nodal data for each time step. This interpolation corresponds to a displacement boundary condition and is used to calculate the electrical potential at the free surface of the sensor.

Experimental data based on air-coupled ultrasound scanning technique can be used for elimination of disadvantages in numerical simulations by developing sensor networks for SHM. In combination with a transfer matrix method (TM) a three-dimensional displacement of specimen surface for complex composites can be calculated. To obtain the sensor signal a surface-bonded sensor is modeled by an isogeometric finite element approach. A good accordance is found between calculated virtual sensor signal and its experimental verification.

Some deviations between calculated signal and its experimental verification are mainly justified by different spectral transfer functions between wave field scanning technique and signal recording of applied sensors. Furthermore, sensor influence on wave propagation is neglected in the presented method.

In this paper, the principle of virtual sensors is applied on anisotropic multilayered lamina by using isogeometric finite elements for piezoelectric sensors. This enables any sensor dimension, layout and position on complex composites. Furthermore a bonding layer between specimen and sensor is considered. The method allows a detailed analysis of sensor behavior on a specimen surface and the design and optimization of entire sensor networks for SHM.