Smart structures health monitoring using artificial neural network

This paper presents a non-model based technique to detect and locate structural damage with the use of artificial neural networks. This method utilizes high frequency structural excitation (typically greater than 30 kHz) through a surface-bonded piezoelectric sensor/actuator to detect changes in structural point impedance due to the presence of damage. Two sets of artificial neural networks were developed in order to detect, locate and characterize structural damage by examining changes in the measured impedance curves. A simulation beam model was developed to verify the proposed method. An experiment was successfully performed in detecting damage on a 4-bay structure with bolted-joints, where the bolts were progressively released.