Diagnosis of normal and abnormal operations of induction motors in ASDs by a coupled finite element-network technique
Abstract
Non-invasive diagnosis of abnormal/faulty rotor conditions of broken bars, broken end-ring connectors as well as static and dynamic airgap eccentricities in squirrel-cage induction motors in Adjustable Speed Drives (ASDs) is a well-known problem that still warrants further investigations. In this regard, there are several important issues that need to be addressed with respect to improvement of the reliability of a drive's condition monitoring and diagnostics. One of these issues is that at present an historical record of performance of a drive or motor is required to detect an increase in the severity of these abnormal conditions. In recognition of these facts, the possibility of future application of numerical model-based predictive techniques, which have the potential for enhanced accuracy and consistency, is increasingly becoming attractive and desirable. Analyses and diagnoses of these various abnormal conditions by use of present techniques as reported in the literature have been restricted, by and large, to the identification of the fundamental component and its sidebands in the frequency spectra of the motor time-domain line current waveforms. Relying on the identification of these sideband frequency components to detect these various abnormalities leaves something to be desired with regard to specific diagnosis of, and differentiation between, broken bars/connectors and airgap eccentricities. A review of present model-based predictive techniques reported in the literature indicates that more attention and emphasis have been directed almost exclusively to the studies of effects of these various abnormalities on the frequency contents of motor line current signatures. To the knowledge of this author, very little attention, if any, has been given to the need for rigorous and detailed representation of important second-order effects, such as electrical and magnetic unbalances and their consequent effects on motor parameters, that inherently occur in these machines as a result of these abnormalities. Such phenomena need to be rigorously incorporated into model-based predictive techniques because they have serious implications on motor diagnostics. Accordingly, in this dissertation a new numerical model is presented that addresses the majority of the issues elucidated above, for rigorous analysis and more comprehensive non-invasive model-based predictive diagnosis of the above-mentioned rotor abnormalities. The potential application of this modeling technique reported here as a powerful diagnostic tool for database generation in identifying the occurrence of, and differentiating between, various faults by monitoring several frequency components in the frequency spectra of measurable and/or computable motor input/output quantities are thoroughly demonstrated via computer simulations in this dissertation. (Abstract shortened by UMI.)
This paper has been withdrawn.