Comparison of Random Forest and Neural Network Framework for Prediction of Fatigue Crack Growth Rate in Nickel Superalloys

2023/09/24 by Pratoori, Raghunandan
#Disordered Systems and Neural Networks (cond-mat.dis-nn) #FOS: Physical sciences

paper · doi:10.48550/arxiv.2309.13534

Abstract

The rate of fatigue crack growth in Nickle superalloys is a critical factor of safety in the aerospace industry. A machine learning approach is chosen to predict the fatigue crack growth rate as a function of the material composition, material properties and environmental conditions. Random forests and neural network frameworks are used to develop two different models and compare the two results. Both the frameworks give good predictions with r² of 0.9687 for random forest and 0.9831 for neural network.

Related

• Predicting Crack Growth and Fatigue Life with Surrogate Models
• Machine learning for predicting fatigue properties of additively manufactured materials
• SIFBench: An Extensive Benchmark for Fatigue Analysis
• Simplified ResNet approach for data driven prediction of\n microstructure-fatigue relationship
• Neural Network and Particle Filtering: A Hybrid Framework for Crack Propagation Prediction