In the era of the 4th industrial revolution of big data, artificial intelligence (AI) is widely used in each and every field of composite materials which includes design and analysis, material storage, manufacturing, non-destructive testing, structural health monitoring (SHM) and prognostics of its remaining useful life, material state (MS) and damage modes. While these AI models are rapidly developed and integrated into the industrial internet of things to keep track of the health of a composite material from its birth to death, these integrations remain uncertain for prognostics without the certainty of its previous MS. This article is a comprehensive review of the AI models being developed over the past few decades in the field of SHM and prognostics health management of polymer matrix composites. It further analyzes the real gaps between these developments and the nature of uncertainty of these methods. Finally, the pipeline for the real-time prognostics from birth to death, hybrid approaches, uncertainty quantification of data-driven and physics-based systems, and its reliability standards to such complex advanced composite materials are discussed. This paper will be focused as a basic guide for researchers implementing AI in composites for diagnosis, prognosis, and control.

No abstract

In recent years, there has been a widespread growth in the application of composite materials particularly in the Aerospace and Automotive sectors. This is because composite structures are generally comparatively light in weight and provide corrosion and wear resistance as compared to metals or ceramics. Due to the strict fail-safe philosophy of the aerospace industry, the certification approach for current practice in joining composite materials is to thicken the joining areas and to use numerous fasteners which in turn increases the weight and stress concentrations in the structure. The use of adhesive bonding can improve the stress distribution between the composite materials / dissimilar materials and can contribute to a lighter structure. However, there much investigation is yet to be done in this discipline to predict the bond strength and performance using non-destructive evaluation methods. This paper will focus on an approach to study the mechanical as well as the dielectric properties of an adhesive bond. The dielectric testing is done by using Broadband Dielectric Spectroscopy (BbDS), wherein the dielectric characteristics of the material are analyzed in a wide frequency spectrum. The data obtained by this technique are used to demonstrate the charge transport, the combined dipolar fluctuation, and the effects of polarization occurring between the boundaries of materials. The continuous modifications of the dielectric spectra are due to the changes in the electrical and structural interactions between the particles, shapes, and orientations of the constituent phases of the morphological structure of the material system. Information about the morphologies, impurities/contamination or interaction of the dissimilar surfaces of the pristine bond can be obtained from the initial BbDS properties. The dielectric properties for adhesively bonded composites with different surface adhesion properties have shown promising evidence of predicting the final mechanical performance of the bonded material system. The success and limitations of this approach will be discussed, and needs for continued investigation identified

Composite materials, by nature, are universally dielectric. The distribution of the phases, including voids and cracks, has a major influence on the dielectric properties of the composite materials. The dielectric relaxation behavior measured by Broadband Dielectric Spectroscopy (BbDS) is often caused by interfacial polarization, which is known as Maxwell-Wagner-Sillars polarization that develops because of the heterogeneity of the composite materials. A prominent mechanism in the low frequency range is driven by charge accumulation at the interphases between different constituent phases. In our previous work, we observed in-situ changes in dielectric behavior during static tensile testing, and also studied the effects of applied mechanical and ambient environments on composite material damage states based on the evaluation of dielectric spectral analysis parameters. In the present work, a two dimensional conformal computational model was developed using a COMSOL TM multi-physics module to interpret the effective dielectric behavior of the resulting composite as a function of applied frequency spectra, especially the effects of volume fraction, the distribution of the defects inside of the material volume, and the influence of the permittivity and Ohmic conductivity of the host materials and defects.

The principal objective of this study is to employ non-destructive broadband dielectric spectroscopy/impedance spectroscopy and machine learning techniques to estimate the moisture content in FRP composites under hygrothermal aging. Here, classification and regression machine learning models that can accurately predict the current moisture saturation state are developed using the frequency domain dielectric response of the composite, in conjunction with the time domain hygrothermal aging effect. First, to categorize the composites based on the present state of the absorbed moisture supervised classification learning models (i.e., quadratic discriminant analysis (QDA), support vector machine (SVM), and artificial neural network-based multilayer perceptron (MLP) classifier) have been developed. Later, to accurately estimate the relative moisture absorption from the dielectric data, supervised regression models (i.e., multiple linear regression (MLR), decision tree regression (DTR), and multi-layer perceptron (MLP) regression) have been developed, which can effectively estimate the relative moisture absorption from the dielectric response of the material with an R2 value greater than 0.95. The physics behind the hygrothermal aging of the composites has then been interpreted by comparing the model attributes to see which characteristics most strongly influence the predictions.