MACHINE-LEARNING-DRIVEN HEALTH MONITORING DIAGNOSTICS FOCUSED ON COMPOSITE STRUCTURES UTILIZING SMART LAYERWISE SPECTRAL ELEMENTS

Abstract

The present study discusses the value of a direct interaction between Structural Health Monitoring applications utilizing active piezoelectric sensors and Machine Learning algorithms in order to develop well established Artificial Intelligence driven diagnostics' tools for composite structures.More specifically, a high-fidelity Time Domain Spectral Finite Element model is developed incorporating physically modeled piezoelectric actuators and sensors as well as mixed order (linear and non-linear) Layerwise Mechanics to simulate efficiently the actual composite structure and all the respective failure modes (fiber, matrix failure and delamination) that could emerge within a composite material .The authors consequently planned a series of damage scenarios which could appear through the thickness of a composite laminate combining both intra-and inter-laminar damage.Afterwards, they simulated the respective virtual experiments of pitch and catch technique corresponding to the aforementioned scenarios by applying an actuating Guassian pulse and acquiring the respective responses of three sensors.Part of these data were then used as input to train different machine learning models predicting the composite material damage and classifying the damage type (output).We examine different approaches for representing the data to feed these machine learning models and present the very promising first findings regarding the effectiveness of the prediction, based on a cross-validation process on the generated simulation data.