Early detection, diagnosis, and even prognosis of a crack in the shaft of a rotor system has been a field of challenging ongoing research for the safe and economic operation of the machinery containing them. This study proposes an embedded modeling methodology for identifying the local stiffness of the shaft in a rotor system from the lateral vibrations in the horizontal and vertical directions. An embedded model integrating a dynamic model of the rotor system and the unknown local stiffness in the form of a neural network is established. A solution method is then used to find the stiffness of the shaft that minimizes the discrepancy between the model output and the measured output. Subsequently, a method is then used to find the location and size of the crack along the shaft using a crack model. Simulated studies were conducted to evaluate if the stiffness of the shaft of a Jeffcott rotor system can be estimated, and to determine if the location can be uniquely identified.
展开▼