A pre-stressed grouted anchor, or simply called “tieback”, transfers tensile forces to a load-bearing ground layer. It consists of a steel tension member anchored with grouted cement and prestressed via an anchor head to secure a structure. Over time, the prestressing force may change due to factors like relaxation, corrosion, or load redistribution. This research uses vibration responses to impulse loads to assess the actual anchor forces. Experiments were conducted on a lab model, measuring vibration responses with accelerometers after impacts from an impulse hammer. Initially, neural networks with supervised learning identified the forces, requiring known prestressing forces from training samples. However, acquiring such data is challenging in practice. To address this issue, the usage of Physics-Informed Neural Networks (PINNs) is outlined. PINNs are designed to integrate mechanical models with measurement data. They can incorporate system properties like mass, stiffness, and damping through equations of motion, compensating for limited training data. This research marks the first application of PINNs to tieback anchors.