FIZIKANI O`RGATUVCHI NEYRON TARMOQLAR (PINN) YORDAMIDA ELASTIKLIK TENGLAMALARINI YECHISH

Yunusova G. A. & Jo`rayev L. G.


Annotatsiya

Ushbu maqolada fizikani o‘rganuvchi neyron tarmoqlar (Physics-Informed Neural Networks — PINN) yordamida elastiklik tenglamalarini yechishning zamonaviy yondashuvi batafsil yoritiladi. Chiziqli elastiklik nazariyasining asosiy tenglamalari, PINN arxitekturasi, yo‘qotish funksiyasi, avtomatik differensiallash (AutoDiff) asosidagi derivativlarning hisoblanishi va amaliy misollar sifatida 2D elastik plastinka va 3D strukturalar masalalari ko‘rib chiqiladi. PINN va an’anaviy сhekli elementlar metodi (FEM) natijalari solishtirilib, PINNning afzallik va kamchiliklari tahlil qilinadi.


Adabiyotlar

  1. Raissi M., Perdikaris P., Karniadakis G.E. Physics-Informed Neural Networks // Journal of Computational Physics. – 2019.
  2. Haghighat E., Juanes R. A Physics-Informed Neural Network Framework for Solving Elasticity Problems. – 2021.
  3. Leake C. 3D PINN Models for Solid Mechanics // Computer Methods in Applied Mechanics. – 2022.
  4. NVIDIA Modulus Documentation. NVIDIA Corporation. – 2023.
  5. Karniadakis G. Deep Energy Method for PDE Solving. Springer. – 2024.
  6. Sadd M.H. Elasticity: Theory, Applications, and Numerics. — Academic Press, 2014.
  7. Zienkiewicz O.C., Taylor R.L., Zhu J.Z. The Finite Element Method: Its Basis and Fundamentals. — Elsevier, 2013