| description abstract | This study addresses the challenge of detecting operational and environmental changes, termed linear damages, in building structures over their lifespan. Identification of these damages is crucial for enhancing serviceability and averting sudden disasters. However, the intricate nature of uncovering concealed changes results in demanding and time-intensive computations, posing a significant computational predicament for related algorithms. Moreover, structures are often exposed to diverse environmental noise, necessitating the development of a robust algorithm capable of effectively identifying subtly hidden damages amid varying noisy conditions with high accuracy and low time consumption. This research introduces a robust and expedited signal-based algorithm, comprising three key components: processing, feature selection, and classification. Multiresolution analysis through discrete wavelet transform is employed for processing, generating diverse features alongside several statistical indices. The grey wolf optimization algorithm is utilized for feature selection, yielding optimal features. This method not only ensures commendable performance under noisy circumstances compared with optimization algorithms such as particle swan optimization and genetic algorithms, as well as common feature extraction methods such as principal component analysis, it also accelerates computation speed by over four times compared with alternative feature-selection techniques such as ReliefF. Lastly, a supervised classification algorithm is integrated to discern distinct predefined scenarios. The efficacy of the proposed algorithm was validated using a comprehensive case study encompassing nine representative scenarios of operational and environmental damages. Incorporating four levels of noise to emulate real-world variations, the algorithm achieved compelling average accuracies of approximately 96%, 93%, 95%, 91.5%, and 89% at original data and signal-to-noise ratios (SNRs) of 10, 5, 1, and 0.5 dB, respectively. | |
