W.S.G. Isaac Newton

Reden B.V. is a company, located in Hengelo (Overijssel), which provides consultancy for engineering problems. The primary focus is on modelling and simulations. Simulations are now often also used in the operational phase of a product. For example in a digital twin to create insight into the state of the system. However, when simulation models are used in a digital twin, the constraints on the simulation time usually increase. In this internship, the objective is to reduce the simulation time of a physics-based model by using a physical-informed neural network (PINN). This model consists of numerous neurons and hidden layers, which can be trained by labelled data and the differential equations governing the problem. The benefit of adding differential equations to the training is that even with limited or noisy data, the PINN model can find an accurate prediction of the exact solution. Before the internship, the knowledge in the domain of neural networks and PINN was very limited. During the internship insight into the possibilities and the limitations in this domain was created and disseminated.

The company has a very welcoming team, who are always available to help when needed, this made the internship enjoyable and a lot has been learned.