Ahmed Mohamed – Mechanical Engineering

Structural Health Prognoisis of Mechanoluminescent Composites using Machine Learning Algorithms

“Smart material particulates impart their multifunctional properties to polymer composites and are referred to as smart particulate polymer composites (SPC). These composites undergo a slow and continuous degradation of its structural strength over millions of cycles due to damage at the interface between the particulate phase and matrix throughout its performance lifetime. As a result of this continuous degradation of mechanical strength, a load much smaller than the design strength of the material leads to abrupt failure. The economic advantages of light-weighting automotive and aircraft component using polymer composites is lost when expensive periodic maintenance becomes mandatory to avoid catastrophic failure. Hence, structural health prognosis (SHP) of SPC is essential in critical load bearing applications structural applications. Structural health prognosis (SHP) utilizing machine learning and deep learning techniques, such as Neural Networks (NN), recently emerged as an efficient tool to predict the health state of a structure in real-time.
A neural network is a data analytical multivariate mapping tool that are trained to inversely relate measured outputs to given inputs with high accuracies.

In this project, a Neural Network-based health prognosis algorithm is constructed by correlating light intensity data to material deterioration, which are acquired by mechanically testing six samples of mechanoluminescent composities. Mechanoluminescence is the phenomenon of light emission from organic/inorganic materials due to mechanical stimuli. This underlying relationship between light intensity and material degradation is developed in the frequency domain by performing the Fast Fourier Transform (FFT) on the measured light intensity data as the composite progressively ages and classifies the light amplitude spectrum as a function of its frequencies and age. The FFT spectrum at different ages or health state encodes vital information regarding structural degradation in the form of changes in amplitude at various constitutive frequencies. Since SHM is basically a pattern recognition problem at given inputs, the large amount of FFT spectrums at different ages and frequencies is fed as inputs to the NN model to identify the underlying pattern and hence map the inverse relationship between measured light intensity and material deterioration. The NN algorithm was trained and tested yielding high accuracies in predicting material deterioration in real-time.”

Leave a Reply

Your email address will not be published. Required fields are marked *