Feature extraction approach insensitive to temperature variations for impedance-based structural health monitoring

Structural health monitoring (SHM) based on the electromechanical impedance (EMI) has been pointed out as a promising method for detecting structural damage. However, practical problems such as the effects of temperature variation on the electrical impedance signatures of piezoelectric transducers have made it difficult to effectively apply this method of detecting damage in real structures. Therefore, in order to contribute to the effective application of the EMI method in real structures, this study presents a new feature extraction approach insensitive to temperature variations. The proposed method is based on the Akaike statistical criterion algorithm, which extracts the number of significant resonance peaks from the electrical impedance signatures. Tests were performed on an aluminium bar with different damage sizes and under different temperatures. The experimental results indicate that the proposed method is capable of detecting and quantifying structural damage in environments under temperature variation.