07/2021 – 10/2021
Predictions of movement can have many different applications, e.g. to predict the effect of a new shoe during the design process. Specifically, we have shown that we can predict energy improvements of novel running shoe materials . However, this research project used a simple ground contact model, which only allows for investigations into a limited amount of parameters. However, we would like to predict more detailed aspects of the running shoe. To do so, we would need to combine a finite element model of the shoe with the trajectory optimization framework. However, a direct combination would be computationally challenging .
Instead, a computationally efficient approach would use a surrogate models, which has been used to create knee force models . Commonly a model order reduction model is used . However, the disadvantage of a model order reduction model is that it does not capture the full finite element model and thus information is lost. Instead, neural networks can approximate theoretically any function without a loss of information. Therefore, we would like to investigate if a neural network can be used to create a surrogate model of a finite element model of a running shoe.
Therefore, the goal of this thesis is to create a surrogate model for a finite element model of a shoe using neural networks. This finite element model is used to analyse the forces and deformations in the shoe. Training data will be available, and the goal is to compare different network architectures and find the possible level of accuracy that can be achieved. Furthermore, we will investigate how well the surrogate model is able to predict unseen load cases.
 Dorschky, Eva et al.: Optimal control simulation predicts effects of midsole materials on energy cost of running. Computer methods in biomechanics and biomedical engineering, 2019.
 Halloran, Jason P. et al.: Concurrent musculoskeletal dynamics and finite element analysis predicts altered gait patterns to reduce foot tissue loading. Journal of Biomechanics 43, 2010.
 Lin, Yi-Chung, et al.: Two-dimensional surrogate contact modeling for computationally efficient dynamic simulation of total knee replacements. Journal of Biomechanical Engineering 131, 2009.
 Qu, Zu-Qing. Model Order Reduction Techniques with Applications in Finite Element Analysis: With Applications in Finite Element Analysis. Springer Science & Business Media, 2004..