چکیده مقاله
In the present study, the Terzaghi consolidation phenomenon has been modeled and analyzed using Physics Informed Neural Networks PINNs Initially, to validate the approach, the results obtained from PINNS were compared with analytical solutions, and the efficacy of the PINN method for modeling soil consolidation was evaluated Subsequently, the effect of the number of neurons per layer in the neural network on the accuracy of the outputs was assessed and analyzed Four different architectures with 3, 6, 9, and 12 neurons per layer were modeled, and the variations in fluid pressure, as the output of the neural network, were evaluated For all analyses, the dimensionless relative error between the predicted values and the exact solution was calculated and interpreted The modeling results of this study indicate that a neural network with 9 neurons per layer provides acceptable results Since the present study was modeled using dimensionless governing equations, the outputs of this research can be utilized by other researchers as well
کلیدواژهها
نویسندگان
Mohammad Bahrami
PhD Student, School of Civil Engineering, College of Engineering, University of Tehran, Tehran, Iran
Reza Rafiee-Dehkharghani
Assistant Professor, School of Civil Engineering, College of Engineering, University of Tehran, Tehran, Iran
شیوه ارجاع
Bahrami, Mohammad and Rafiee-Dehkharghani, Reza,1404,Effect of the Number of Neurons per Layer on the Consolidation Behavior in PINN Modeling,14th International Congress on Civil Engineering,Tehran
ارائهشده در
مجموعه مقالات چهاردهمین کنگره بین المللی مهندسی عمران29 مهر 1404 · تهران