Z-ENG: Optimization Techniques to tune Soil Models
Soil models are widely used in the field of agriculture and civil engineering. Various soil simulation models exist in the literature and are also implemented. Generally speaking, the main goal of the simulation is to provide a realistic digital twin of the current problem, so that experiments can be conducted in a cost optimal way. As models typically generalize the actual problem, a model error is typically present. On the other hand, the actual parameters of the real world problem are unknown or cannot be measured. The goal of this project is to find the parameter setting of the soil model that leads to the minimum of the modeling error, which minimum should be the global minimum. The difficulty of the task is that the error surface is typically not smooth at all, which makes it hard to find a global optimum.
The student will work on the implementation of optimization algorithms and apply the algorithms to the actual problem, to fit the soil model. Both local and global optimization techniques are to be implemented and evaluated.