Condition assessment of concrete plates using impulse-response test with affinity propagation and homoscedasticity. (1st October 2022)

• Record Type: Journal Article
• Authors: Sajid, Sikandar; Chouinard, Luc; Carino, Nicholas
• Is Part Of: Mechanical systems and signal processing. Volume 178(2022)
• Publication Date: 2022-10-01
• Subjects: Debonding -- Impulse-response test -- Delamination -- Honeycombing -- Homoscedasticity -- Affinity propagation; Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2022.109289
• Languages: English
• ISSNs: 0888-3270

Abstract

Abstract: This article presents a procedure based on the impulse-response test and the affinity propagation algorithm to detect delamination, honeycomb, and debonding in concrete slabs. Frequency response functions (FRFs) generated with the impulse-response test according to the protocols of ASTM C1740 on a grid pattern are combined to define a feature space matrix. The affinity propagation (AP) algorithm is used to cluster the FRFs sharing similar features. Clusters are shown to correspond to either intact locations or defects and cluster membership can be used to locate defects in the slab. The effectiveness of the algorithm for defect delineation depends on the optimal number of exemplars and some objective criteria to categorize each cluster according to the severity of defects. To achieve this objective, an upper bound for the number of exemplars is proposed based on the number of principal components explaining most of the percent variance in the feature matrix for optimal defect detection and the variance of the FRF is proposed and validated as a criterion for ranking the severity of defects. Homogeneity of variance testing is used for merging exemplars that exhibit similar levels of variability to address potential overfitting associated with the AP algorithm. Two fully supported slabs with free edges and having different simulated defects, and a slab cast on a rigid insulation board with delamination and honeycomb are used to validate the proposed procedure. The proposed procedure is shown to efficiently detect shallow delaminations and debonding. The deep delamination and honeycomb in the slabs were delineated but to a lesser extent compared with the shallow delamination.