Experimental and computational vibration analysis for diagnosing the defects in high performance composite structures using machine learning approach

Delamination in laminated structures is a concern in high-performance structural applications, which challenges the latest non-destructive testing techniques. This study assesses the delamination damage in the glass fiber-reinforced laminated composite structures using structural health monitoring techniques. Glass fiber-reinforced rectangular laminate composite plates with and without delamination were considered to obtain the forced vibration response using an in-house developed finite element model. The damage was diagnosed in the laminated composite using machine learning algorithms through statistical information extracted from the forced vibration response. Using an attribute evaluator, the features that made the greatest contribution were identified from the extracted features. The selected features were further classified using machine learning algorithms, such as decision tree, random forest, naive Bayes, and Bayes net algorithms, to diagnose the damage in the laminated structure. The decision tree method was found to be a computationally effective model in diagnosing the delamination of the composite structure. The effectiveness of the finite element model was further validated with the experimental results, obtained from modal analysis using fabricated laminated and delaminated composite plates. Our proposed model showed 98.5% accuracy in diagnosing the damage in the fabricated composite structure. Hence, this research work motivates the development of online prognostic and health monitoring modules for detecting early damage to prevent catastrophic failures of structures.