Fault Diagnostics of Rotating Machinery

This research extends the concept of [Lehrasab, 1999] to the domain of rotating machines. It aims at developing novel fault diagnostics techniques for the motor or rotating machines for diagnostics that is costeffective and easier to deploy from industrial perspective. Normally the encoders data is only used to measure speed, however this paper attempts to use the same data normally considered insufficient for mathematical modeling, in creating a set of model based residuals. The velocity measurements based on the timing between the hall effect sensors placed around the motor in form of a mini-plastic belt is accurately measured with a microprocessor without involving expensive A/D. Conventional schemes usually deploy the spectral analysis that requires expensive sensors and powerful data processing engines. This scheme employs the conventional encoder and with a very limited floating point implementations. The complete model based residual generator can be implemented that can relate the rotational jitters and mechanical faults and with further analysis it can be linked with a set of transformations from spectral domain to the jitter domain. The paper presents a novel model based residual generator for shaft Encoder of a motor based on its jitter model. The mechanical errors of the system are translated into variations in the angular velocity. The residual generator presented in this paper is sensitive to the resolution of jitters of 4us to cross 3s thresholds.