Font Size: 

In this paper we ultilize the emerging paradigm of physics-informed neural networks (PINNs) for inverse problems in elastic rods. Specifically, we sucessfully employ mesh-free PINNs for obtaining the distribution of longitudinal speed in elastic rods from the displacement measurement at one end. The methodology is fully verified by numerical simulations based on the Finite Element Method. The implementation of physics-informed machine learning for inverse wave scattering problems enables the identification of material properties in many engineering structures.

inverse problems, elastic rods, physic-informed neural networks.