| Researchers: |
Ing. Bulko Roman, PhD., Ing. Daniel Ľuboš, PhD., prof. Ing. Kovařík Karel, CSc., Ing. Masarovičová Soňa, PhD., Ing. Gago Filip, PhD., Ing. Mihálik Ján |
| The project focuses on the design and implementation of soil constitutive models based on the neural network paradigm. The neural constitutive model should better describe the highly nonlinear behavior of the soil and implicitly describe some aspects of the response to mechanical stress, which is often not described even when using advanced soil constitutive models. The project involves the design of an application framework for the implementation of the neural network architecture and its learning process. Advanced constitutive models are rarely used in geotechnical practice, mainly due to the complexity of the model formulation and the large number of input parameters, which in some cases do not represent a direct physical characteristic, and their determination is quite problematic and non-intuitive. The goal is to implement a neural constitutive model capable of competing with advanced soil constitutive models, and its subsequent integration within the framework of numerical analyses using the finite element method. |
