Research On Rotor Broken Bar Fault Diagnosis Of Induction Motor

Induction motors are widely used in the industrial and agricultural production because of its simple structure, low price, high reliability and convenient maintenance. With the rapid development of the modern industrial system, the capacity of a single motor is keeping increasing and the load is also becoming more complicated now. A motor failure not only can result in damage to the motor, but also can result in unscheduled machine downtime and the shutdown of a production line, which will cause heavy financial losses and catastrophic failure. Statistical studies have shown that the rotor broken bar fault, which account for nearly 10% of total induction motor failures, is the most familiar fault for induction motors. Consequently, researches on rotor broken bar fault diagnosis able to detect this kind of fault at an early stage and also to allow for carefully planed repair actions are of great theoretical significance and socioeconomic benefits, which has turned to be the hotspot of research for scholars in the world. This paper presents several methods to diagnose rotor broken bar fault based on the electromagnetic torque and the instantaneous active and reactive power, which are highly sensitive to any asymmetrical operating condition. These parameters derive from multiple input voltage and current signals and can offer potential advantages over the one phase current of the induction motor.The floating threshold filter technique based on the frequency band decomposition of the wavelet packet transformation, which is superior to the Fourier spectrum analyze method for non-stationary signal denoising, is presented. An arbitrary signal and its Hilbert transformation constitute an orthonormal pair. Based on this idea, the principle of DC elimination, which is available for an arbitrary signal with random non-integral periods, is proposed by the double Hilbert transformation. The Hilbert-Huang transformation which combines the empirical mode decomposition and the associated Hilbert spectral analysis are presented. The variable amplitude and frequency of the intrinsic mode function, is a breakthrough of the constant amplitude and fixed frequency Fourier series expansion. The Hilbert-Huang transformation is auto adaptive for the non-stationary signal analyzing.The expression and frequency variety law of the rotor broken bar fault feature component in the startup electromagnetic torque signal are analyzed in detail. On the one hand, the startup electromagnetic torque signal is decomposed into a number of intrinsic mode function components based on the principle of empirical mode decomposition. The rotor broken bar fault feature component can be extracted based on instantaneous frequency estimation using the intrinsic mode function component which includes rotor broken bar fault information. The relationship between its Hilbert marginal spectrum and the number of adjacent rotor broken bars is given, which provides a new way to evaluate the rotor fault severity. On the other hand, the frequency variety law of the rotor broken bar fault feature component in the startup electromagnetic torque signal can be extracted based on the wavelet ridge algorithm. The fault feature torque magnitude variety law can be reflected by the waveletcoefficients modulus of the fault characteristic ridge because of the energy conservation of the wavelet transformation. The ridge-wavelet energy spectrum is thus defined based on the principle. Using the ridge-wavelet energy spectrum of the fault characteristic torque as the rotor fault indicator, the second expression of the fault severity index is given.With the positive sequence fundamental voltage components and its minus Hilbert transformation as elements, the new PQ transformation matrix is constructed and the rotor broken bar fault diagnosis method based on the double PQ transformation is presented. The active and reactive power are both constant for the healthy motor, therefore, they correspond to a dot on the PQ reference coordinate. However, because of the additional fault feature component in the active and reactive power of the rotor broken bar motor, they will correspond to an ellipse. By distinguishing of these two different patterns, the rotor broken bar fault can be detected. Using the major radius of the ellipse as the fault indicator and the distance between the point of no-load condition and the center of the ellipse on the PQ reference coordinates as its normalization value, the fault severity factor which is approximately independent of load level and mechanical inertia value of the induction motors is given.The current sideband components due to rotor broken bar fault and load fluctuation are deduced in detail. Performing PQ transformation on them respectively, the active and reactive power components corresponding to them can thus be obtained. Using the magnitude of the feature power components directly derived from rotor fault as the rotor fault indicator, a new fault severity index which is completely independent of load level and mechanical inertia is given. Based on the different orientation of the feature ellipse of the rotor faulted and load fluctuated motor, a method to reliably discriminate between rotor faults and load fluctuation is presented.The main results in this paper, which indicate not only the superior theoretical significance but also the high engineering practical value, have been demonstrated by experiments. The developed rotor broken bar fault diagnosis system, which has met the design objective and has a delightful foreground, has been tested in the lab and put to use in the industry plants.