Damage detection in composite structures using piezoelectric materials (and neural net)

Publisher Website
Google Scholar
Abstract
Damage in composite structures is detected by embedded piezoceramic sensors. Modal analysis is carried out using piezoceramic patches as both sensors and actuators on structure members. Using piezoceramic actuators, the member is excited at different sinusoidal frequencies, and, using piezoceramic sensors, the response of the member is measured. The advantage of using active materials for system identification is that the condition of the structure can be continuously monitored, and, by using an integrated microprocessor, the sensor output can be continuously evaluated. A back-propagation neural network has been trained with the frequencies of the first five modes obtained from modal-analysis data from piezoceramic sensors in both damaged and healthy composite beams. The effectiveness of neural networks in determining the location and size of any delamination is discussed.