LAUSR

Abstract Casing vibration is usually applied for diagnosing faults of rotor-system in aero-engine, while the transmission mechanism of the vibration from the inner source to the outer casing is rarely discussed and the vibration propagation characteristics have not been quantitatively evaluated. This work presents a data-driven spatiotemporal graphical modeling approach, spatiotemporal pattern network (STPN), to estimate the vibration propagation characteristics by exploring the causal dependency between the measurements. The proposed STPN model with information based metric is further applied to ascertain the most sensitive position on the casing for diagnosing the faults of the rotor system. Numerical simulation and experiments on a test rig are carried out to validate the proposed approach. The results show that the proposed approach is capable of determining the vibration transmission characteristics from the inner rotor system to the outer casing and can be applied in sensitivity analysis to ascertain the best sensor location on the casing.