Abstract:Accurately predicting the initial peak crushing force (IPCF) of the CFRP/metal hybrid structure is of great significance for the high-efficiency energy absorption of energy-absorbing elements and ordered deformation of combined energy-absorbing devices. Due to the significant differences in material properties between CFRP and metal and the high complexity of cross-sectional shape, a combined method of machine learning and mechanism analysis is proposed for predicting the IPCF of CFRP/Al hybrid energy-absorbing tube. First, a high-precision finite element model validated by the axial and oblique experiments is conducted to generate the dataset for predictive analysis. Second, a sample space through the design of experiment (DOE) is established for data fitting between structural parameters and IPCF based on traditional response surface model (RSM). It is found that traditional RSM method can successfully predict the IPCF of the single-cell hybrid tube, but cannot effectively predict the IPCF of the multi-cell tube due to the strong coupling effect between inner and outer plates. To address the issue of the lack of accurate correspondence between structural parameters and IPCFs in multi-cell hybrid tube, the indicator of relative density is proposed to characterize complex cross-sectional features. Surprisingly, there is a strong correlation between relative density and IPCF where multi-cell tubes with the same relative density have almost the same IPCF. Third, different artificial neural networks (ANN) are trained by a pre-simulated IPCF database of CFRP/Al hybrid energy-absorbing tube depending on the parameter characterization scheme based on mechanism analysis. In order to enhance the network's performance, the optimal configuration parameters of the neural network, including the number of layers and nodes, are ascertained through a combination of K-fold cross-validation and the Grid Search method. As a result, compared with the traditional RSM and Kriging model, the IPCF of CFRP/Al hybrid energy-absorbing tube can be predicted in small samples with high accuracy using the optimal ANN model. Finally, Through the comparative analysis of the IPCF of hybrid tubes with reported multi-cell configurations, it is found that the prediction model proposed in this paper has good generalization ability, which can be used to predict more complex types of multi-cell tube configurations. The present study offers important suggestions for the overall control of multi-materials hybrid energy-absorbing tube and the ordered deformation of combined energy-absorbing devices.

Accurate Loss Prediction of Realistic Hollow-core Anti-resonant Fibers Using Machine Learning

Random Design Variations of Hollow-core Anti-resonant Fibers: A Monte-Carlo Study

Machine Learning and Materials Informatics Approaches for Predicting Transverse Mechanical Properties of Unidirectional CFRP Composites with Microvoids

Analytical assessment of the total loss in tubular hollow-core fibers during their fabrication process

Recent Progress in Low-Loss Hollow-Core Anti-Resonant Fibers and Their Applications

Simulation of the Structural Parameters of Anti-resonant Hollow-core Photonic Crystal Fibers

Poor-man's model of hollow-core anti-resonant fibers

Theoretical Analysis and Fabrication of a Nested Antiresonant Hollow-Core Fiber with Low Tight Bend Loss

A machine learning approach for predicting the Johnson-Champoux-Allard parameters of a fibrous porous material

Machine Learning Prediction of Mechanical Properties of Braided-Textile Reinforced Tubular Structures

Design of 2 μm Low-Loss Hollow-Core Anti-Resonant Fibers

Reverse engineering and analysis of microstructure polymer fiber via artificial neural networks: simplifying the design approach

The Research on Large-Mode-Area Anti-Bending, Polarization-Insensitive, and Non-Resonant Optical Fibers

Deep neural network for microstructured polymer fiber modeling

Low bending loss few-mode hollow-core anti-resonant fiber with glass-sheet conjoined nested tubes

Nested antiresonant nodeless hollow core fiber

Honeycomb-based method for generating random fiber distributions of fiber reinforced composites and transverse mechanical properties prediction

A machine learning-enabled prediction of damage properties for fiber-reinforced polymer composites under out-of-plane loading

Predicting optical parameters of nanostructured optical fibers using machine learning algorithms

Data-driven initial peak crushing force prediction of hybrid tubes

Highly Birefringent Anti-Resonant Hollow-Core Fiber With a Low Loss