Machine Learning Applied to Composite Materials

✍ Scribed by Vinod Kushvaha, M. R. Sanjay, Priyanka Madhushri, Suchart Siengchin

• Publisher: Springer
• Year: 2022
• Tongue: English
• Leaves: 202
• Series: Composites Science and Technology
• Category: Library

✦ Synopsis

This book introduces the approach of Machine Learning (ML) based predictive models in the design of composite materials to achieve the required properties for certain applications. ML can learn from existing experimental data obtained from very limited number of experiments and subsequently can be trained to find solutions of the complex non-linear, multi-dimensional functional relationships without any prior assumptions about their nature. In this case the ML models can learn from existing experimental data obtained from (1) composite design based on various properties of the matrix material and fillers/reinforcements (2) material processing during fabrication (3) property relationships. Modelling of these relationships using ML methods significantly reduce the experimental work involved in designing new composites, and therefore offer a new avenue for material design and properties. The book caters to students, academics and researchers who are interested in the field of material composite modelling and design.

✦ Table of Contents

Contents
Applications of Machine Learning in the Field of Polymer Composites
1 Introduction
2 An Overview of the Applications of Machine Learning in Polymer Composites
3 Generally Used Machine Learning Algorithms in Polymer Composites
3.1 Neural Networks
3.2 Logistic Regression
3.3 Gaussian Process
3.4 Support Vector Machine
4 Conclusion and the Future Scope
References
Image Processing and Machine Learning Methods Applied to Additive Manufactured Composites for Defect Detection and Toolpath Reconstruction
1 Introduction
2 Methods
2.1 Sample Preparation
2.2 μCT Scan Image
2.3 Circular Image Dataset Preparation for ML Model Training
2.4 Dataset for Orientation Prediction in Each Layer
2.5 Machine Learning
2.6 The Architectures of the Machine Learning Algorithms
3 2D CNN Result
3.1 1-Dimensional Convolutional Neural Network (1D CNN)
3.2 1D CNN Result
4 Summary
References
AI/ML for Quantification and Calibration of Property Uncertainty in Composites
1 Introduction and Prior Work in Modeling and Uncertainty Quantification
1.1 Representation of Property Variations in Materials
1.2 Sampling with Monte Carlo Methods
1.3 Sampling with Reliability Methods
1.4 Sampling with Spectral Stochastic Finite Element Methods
1.5 Approaches for Micro- and Macro-UQ of Composite Performance
1.6 Machine Learning Methods for UQ
2 Enhancing Sample Efficiency Through Quasi Monte Carlo Simulations
2.1 Application of QMC for the Modeling of Random Fields with KLE
3 UQ Parameter Calibration with Neural Networks
3.1 Karhunen-Loève Expansion
3.2 Computational Model
3.3 Parameter Estimation Using a Neural Network
3.4 Network Architecture
3.5 Experimental
3.6 Neural Network Training
3.7 Optimization of Neural Network Architecture
3.8 Estimation of Experimental Correlation Lengths
4 Concluding Remarks
References
Radial Basis Function-Based Uncertain Low-Velocity Impact Behavior Analysis of Functionally Graded Plates
1 Introduction
2 Mathematical Formulation
3 Radial Basis Function (RBF)
4 RBF-Based Probabilistic Analysis of FGM Plates
5 Results and Discussion
6 Conclusions and Future Perspective
References
Application of Machine Learning in Determining the Mechanical Properties of Materials
1 Introduction
2 Big-Data
3 Machine Learning
4 Tensile Strength
5 Fatigue and Creep
6 Conclusion
References
Machine Learning Prediction for the Mechanical Properties of Lightweight Composite Materials
1 Introduction
2 Background of FRPs
3 Non-destructive Testing for FRPs
4 Overview of Machine Learning
5 Machine-Learning Prediction of Mechanical Properties of FRPs
6 Conclusion
References
Ballistic Performance of Bi-layer Graphene: Artificial Neural Network Based Molecular Dynamics Simulations
1 Introduction
2 Modelling and Simulation
2.1 MD Setup
2.2 Artificial Neural Network
3 Results and Discussion
4 Conclusion and Future Perspective
References
Quantifying the Sensitivity of Input Parameters in an ANN-Based Committee Networks Model for Estimation of Steel Girder Bridge Load-Ratings
1 Introduction
2 Methodology
2.1 Bridge Datasets
2.2 Load Rating
2.3 FE Modeling and Calibration
2.4 Committee Neural Networks (CN)
2.5 Sensitivity Analysis of Network Inputs
3 Discussion of Results
4 Conclusions
Appendix
References
Estimating Axial Load Capacity of Concrete-Filled Double-Skin Steel Tubular Columns of Multiple Shapes Using Genetic Algorithm-Optimized Artificial Neural Networks
1 Introduction
2 FE Modeling and Simulation
2.1 Element Selections
2.2 Constitutive Material Models
2.3 Loading and Boundary Conditions
2.4 FE Modeling Validation
3 Discussion of Results
3.1 Datasets for Neural Network
3.2 ANN Model Predictions
3.3 Hybrid GA-ANN Model Predictions
4 Conclusions
Appendix 1
Appendix 2
References