The integration of Physics-Informed Neural Networks (PINNs) into Prognostics and Health Management (PHM) through digital twin modeling is a rapidly expanding field. In this study, the use of a deep learning approach supported with physical laws is studied and applied to a proportional high bandwidth servo valve, part of a force controlled hydraulic servo system. The ASTIB test bench [4] is used to support this work; an advanced system for the testing of electro mechanical actuators (EMA), powering flight control surfaces and landing gear. The hydraulic force controlled part simulates external force disturbances from wind and guts. Feasibility of such an approach is evaluated by implementation of a general compressibility law of the fluid powering a hydraulic linear actuator (HLA), on the data driven model.