Papers by Author: M. Ghajari

Abstract: A SMART Platform is developed based on sensor readings for Structural Health Monitoring of a stiffened composite panel. The platforms main function is divided into three categories: Passive sensing, Active sensing and Optimal sensor positioning. The platform has self-diagnostic capabilities, i.e. prior to its application the health of the sensors and their connection will be checked to avoid any false alarm. Passive sensing results in impact location and force magnitude detection. Active sensing is performed for damage detection. It results in detecting the damage location and severity. Finally the optimal sensor location can be provided given the number of sensors and probability of detection value. This platform is the first step in applying the developed SHM methodologies to real size structures in service load conditions.

Abstract: A number of small mass and large mass impacts on a sensorised aircraft stiffened panel were numerically simulated. Sensor signals and the contact force history were recorded during each impact. A significant difference was noticed between the small mass and large mass impacts with respect to the contact force. To distinguish between these two types of impacts, the Fast Fourier Transform was performed on the sensor signals and a categorisation criterion was defined. Finally, two separate Artificial Neural Networks were trained to approximate the peak contact force for each type of impact. It was found that the performance of these ANNs were better than a single ANN trained for both small and large mass impacts.

Abstract: In this work, a number of impacts on a composite stiffened panel fitted with piezoceramic sensors were simulated with the finite element (FE) method. During impacts, the contact force history and strains at the sensors were recorded. These data were used to train, validate and test two artificial neural networks (ANN) for the prediction of the impact position and the peak of the impact force. The performance of the network for location detection has been promising but the other network should be further improved to provide acceptable predictions about the peak force.