A New Strategy for Damage Identification in SHM Systems by Exploring Kappa Coefficient

Abstract

Recently, considerable research works have been conducted towards finding fast and accurate pattern classifiers for identifying structural damage when applied to Structural Health Monitoring (SHM) systems. In this way, this paper presents a novel approach for damage identification in SHM systems by proposing the use of the Kappa coefficient as a metric for damage detection in SHM systems. Here, Kappa is also proposed along with PSO-Fuzzy ARTMAP neural network aiming to reduce the amount of attributes, leaving only the most representative features. It is important to highlight that Kappa coefficient analyses the whole confusion matrix instead of using only the principal diagonal as used, for instance, when we compute success rates. Additionally, the Particle Swarm Optimization (PSO) algorithm is used in searching for a set of criteria employees in training damage classifier, in order to maximize the accuracy rate of the classifier. Hence, the Fuzzy ARTMAP Network (FAN) algorithm is used to identify structural damage. The performance of this new strategy is evaluated considering an experimental setup based on the Electromechanical Impedance (EMI) technique, in the time domain, which uses four piezoelectric patches glued onto a unidirectional composite plate. Damage scenarios were simulated by loosening bolts at different positions. The paper discusses the effectiveness of the proposed methodology in light of the experimental results.