Aluminium Plate Surface Defect Detection and CLassification based on Piezoelectric Transducers

This paper investigates a detection technique for monitoring the structural health of an Aluminium plate. Inspection of aluminium metal plate surface enables ease of product maintenance and provides warnings to prevent any catastrophic failure. The proposed diagnostic technique enables the detection of surface defects in aluminium metal plates based on piezoelectric transducers. Firstly, the natural frequency response for the plates was obtained using the Finite Element Method (FEM) of simulation. The natural frequency for each defective plate was compared which shows different frequencies for the different defective plates. Secondly, a piezoelectric actuator was used to produce periodic excitation to the aluminium plate, the mechanical impact response is sensed using a piezoelectric sensor. Finally, the sensor responses were used to classify the defects using the K-nearest neighbours (KNN) classification algorithm which yielded an accuracy of 91.75%.