Time-scale local approach for vibration monitoring

It is difficult to monitor machine condition and diagnose mechanical faults in mechanical equipment which have varied operational modes and whose dynamic signals are nonstationary. The identification of natural frequencies of vibration and modal damping is of fundamental engineering importance. The damage done to a particular structure under an unknown force might be ascribed to the relative damping or the lack thereof and the natural frequencies of a vibrating system. The paper focuses on the design of a fault detection algorithm, which consists of two steps: (1) residual generation at each scale and (2) decision rules. The designed residual measurements are to reflect possible changes of interest in the dynamic signal. They are based on sufficient statistics of the model parameters at each resolution level. The decision rule is based on the generated residual. The following decision mechanisms are implemented: chi-squared test, and Shewhart control chart. The vibrating system that is considered in this research, is the gear dynamic system. The dynamic system considered in this research is a vibrating gear system