Using Artificial Neural Networks to Predict Critical Displacement and Stress Values in the Proximal Femur for Distinct Geometries and Load Cases

• Type: Chapter or Part of a Book
• Date: 2023

Authors

Pais, A (Author) Other; Jorge Lino (Author) FEUP; Belinha, J (Author) Other

Indexing

Scopus - 0 Citations

Other information

Authenticus ID: P-00Z-J41

DOI: 10.1007/978-3-031-44127-1_2

Abstract (EN): In computational mechanics, the finite element method (FEM) is a very common discretization numerical technique. The complexity of numerical applications, however, is rising today. As a result, classic solution methods typically require more processing power and exhibit higher computational costs. To lower the computing cost associated with the numerical analysis, machine learning approaches can be coupled with the FEM and used as surrogate solvers or as a prediction tool. This alternative was examined in order to demonstrate the possibilities of fusing artificial NN with FEM for a biomechanical application. The proximal femur was used as a numerical example. Thus, distinct geometries were generated and to each discretized model different load cases were applied. Then, all the discrete models were analyzed with the FEM, and the initial conditions (geometry and load cases) and the obtained results (displacements and stresses) were organized as input and output data, respectively. The ANN was trained and then its accuracy was verified. It was observed that artificial NN can accurately forecast displacements and stresses while also saving a significant amount of computing time. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023.

Language: English

Type (Professor's evaluation): Scientific

No. of pages: 11