Evaluation of matrix cracking in composite laminates based on anomaly indices. (November 2020)

• Record Type: Journal Article
• Authors: Liu, Xiaofeng liuxfeng; Wang, Bangxin; Ai, Fan; Wei, Daiping; Bo, Lin
• Is Part Of: International journal of fatigue. Volume 140(2020)
• Publication Date: 2020-11
• Subjects: Matrix cracking -- Recursive quantitative analysis -- Anomaly indices -- Feature fusion -- Damage evaluation; Materials -- Fatigue -- Periodicals; Materials -- Fatigue; Periodicals; 620.1122
• DOI: 10.1016/j.ijfatigue.2020.105841
• Languages: English
• ISSNs: 0142-1123

Abstract

Highlights: The RQA features are utilized to characterize matrix cracking in composite laminates. An anomaly-detection method is proposed to detect the early-stage matrix cracking. An anomaly indices is obtained to quantify the severity degree of matrix cracking. The simulation and experiments verified the effectiveness of proposed method. Abstract: The paper aims to develop a Lamb waves-based automated method dealing with early detection and accurate evaluation of matrix cracking in composite laminates. An anomaly-detection method devoted to the quantification of the extent to which a laminate plate deviates from its intact state is developed. In view of the nonlinear and chaotic dynamic properties of laminate plate caused by matrix cracking, recursive quantitative analysis (RQA) of sensed Lamb waves is introduced to characterize matrix cracking in composite laminates. The problem of damage detection is then recast as one-class classification problem in the space spanned by a set of RQA features, with the aim of global differentiation between normal and anomalous observations, respectively related to intact and supposed damaged laminate. To further quantify the damage degree of matrix cracking, an anomaly indices (AI) based on support vector data description (SVDD) is formed by combining a set of RQA features correlated with matrix cracking, thus the operating condition variability is implicitly included in the model through the feature fusion. The results of simulations and experiments verified that the proposed AI is sensitive to matrix cracking and can be used to quantify its severity degree.