IMPACT LOADING ANALYSIS OF PARTICULATE POLYMER COMPOSITES WITH AN EFFICIENT HYBRID MACHINE LEARNING APPROACH

Abstract

The fracture behaviour of particle composites made of polymers under impact loading is predicted in this research using a hybrid machine learning approach dubbed Hybrid Artificial Neural Networks and Random Forest (HANN-RF), with a focus on mode-I fracture. The goal of the study is to create a model for prediction that accurately links input variables to histories of crack initiation, fracture toughness, and the intensity of the stress factor (SIF). A full dataset is created, with inputs for the composites' compositional properties and impact loading scenarios. The HANN-RF model combines a Random Forest (RF) method and an ANN (Artificial Neural Network) in order to improve robustness and accuracy in forecasting. Metrics like MAE, MAPE for short, and accuracy are used in model evaluation. The outcomes show that the HANN-RF technique successfully predicts and forecasts mode-I fracture behaviour, offering insightful information for evaluating the effect on resilience and longevity of particle polymer composites in a variety of applications.

Keywords

• particulate polymer composites
• machine learning
• hybrid artificial neural network
• and random forest (hann-rf)