چكيده به لاتين
The structural health monitoring (SHM) based on a new philosophy in maintenance of the structures as the new generation of Non-destructive tests, which is powered by modern concepts, equipments and technologies is important to be considered.
Ultrasonic guided waves have been widely used for SHM applications thanks to capability of its traveling in long distance, being sensitive to small damages and the ability to propagate in the solid environment with high level of attenuation such as composite materials. New methods based on SHM are more efficient in comparison with the conventional Non-destructive tests, such as ultrasonic test.
The aim of this research is to design and optimize a structural health monitoring system by ultrasonic guided waves and use of the piezoelectric transducers with dual roles of exciters and sensors in order to detect various types of damages in structures.
After dividing GF/RP composite plate specimen into 4 zones based on the network arrangement of the piezoelectric transducers, 40 different paths are considered on the structure to compare intact zone with defected zones via emmited signals.
By changing the role of the exciters and sensor for each transducer attached to the structure, excitation of the transducers is done by the function generator and the ultrasonic guided waves propagate in the plate. For each path with reception of the waves by sensors, the signal relates to each sensor is received by digital oscilloscope and designed software in LabVIEW; so, it was saved in the computer after de-noising.
The extraction of the features from signals is done by the software programming in MATLAB, and via design of different indices and study their changes has been done by advanced signal processing techniques in the joint time-frequency domain, such as wavelet transform. Finally, the possibility of comparing received signals from different zones of the structure were considerably provided to identify and classify three different types of damages (delamination, notch and hole) due to various effects of each damage on the features of ultrasonic guided waves (the fingerprint of any damage on incoming waves).
At the end, a multi layers Perseptron (MLP) neural network which has the extracted signal features as an input was designed and trained with back-propagation algorithm. Accordingly, the designed neural network can function appropriate to identify three types of damages and in addition, prognosis the type of damages based on the designed algorithm. Therefore, this is considered as an innovative method to recognize and classify the possible damages in a composite plate-like structure in an unknown condition as a leading part of a modern SHM system.
