Modeling and Identification for Anomaly detection in Mechanical System

The main objective of the work reported in thismonograph is the fabrication of an experimentalapparatus to detect slowly evolving anomalies (i.e.,decrease in stiffness due to fatigue failure)in complex mechanical systems at an early stage byobserving time series data of the availableultrasonic and performance measuring sensors. Theinformation on evolving anomaly will serve as aninput to the life extending control policy that will,in turn, generate corrective actions to mitigatefatigue crack damage in the specimen structures andthereby extend the remaining life of the systemwithout any significant loss in performance.The monograph presents the design requirements forthe experimental testbed. Physical modeling gives outthe expected resonant frequencies. Finally, systemidentification presents the actual resonantfrequencies achieved from the setup and compares thesame with the theoretical physical model. Themonograph discusses the proximity of the resonantfrequencies achieved through physical modeling andsystem identification and presents conclusions on thesame.