@misc{Klejment_Piotr_Failure,
 author={Klejment, Piotr},
 howpublished={online},
 publisher={Zielona Góra: Oficyna Wydawnicza Uniwersytetu Zielonogórskiego},
 language={eng},
 abstract={In this paper, an automated methodology for predicting the stress state and time to failure of a material during a uniaxial compression test was proposed. It was shown that, based solely on pseudo-acoustic emission, the supervised machine learning algorithm Random Forest can perform predictions with good or very good accuracy. The Coefficient of Determination R2 on the test dataset reached 84% (for axial stress prediction) and 73% (for time to failure prediction). This work was limited to predictions only in numerical modeling using the Discrete Element Method. Cylindrical samples with macroscopic parameters corresponding to four real sandstones were generated. SHapley Additive exPlanations (SHAP) was applied to show what is the contribution of individual features of pseudo-acoustic emission to the algorithm`s output and its predictions.},
 title={Failure Prediction With (pseudo) Acoustic Emission and Supervised Algorithm Random Forest - Case Study of Four Numerical Sandstones},
 type={artykuł},
 keywords={uniaxial compression, numerical modelling, supervised machine learning, random forest, discrete element method},
}