The emerging research on Digital Twins (DTs) for prognostics has been vital to the future of Industry 4.0. These DTs can be utilized to increase the accuracy and efficiency for prognostics and health management (PHM) for complex industrial equipment. This is based on the approximation of the Remaining Useful Life (RUL) which requires an accurate digital twin that represents the dynamics of the physical system. However, the dynamics for the physical system do not remain constant due to structural changes over time. Research by our group on the application of an equation discovery technique known as Model Error Discovery with Interpretability and Data Assimilation (MEDIDA) for DTs has proven feasible for structural identification and model updates. This thesis analyzes the elements of a DT for prognostics through structural identification for an Industrial Emulator (IE) experimental setup. A closed-loop system was determined necessary for structural identification and was used to successfully identify a single nonlinearity and a nonlinearity caused by frictional dynamics of the LuGre friction model with an additional unmeasurable state. These elements are crucial to obtain a DT that replicates the physical system and provides a more accurate RUL calculation for prognostics.