Detection of coil shorting in an induction motor by means of wavelet detectors based on orthogonal Legendre polynomials

This paper presents the application of narrowband detection wavelet filters, which use the Legendre orthogonal polynomial family, in the diagnostics of coil shorting in induction motors. Wavelet orthonormal bases were built with the use of the recurrent relationships generating higher-degree Legendre polynomials. For the scaling function constructed in the form of linear combination of Legendre polynomials of the selected levels, synthesis of the basic wavelet was performed - this mapped the characteristic properties of the signal carrying the information about the nature of the fault. A signal proportional to the machine axial flux was used for the analysis - this enabled the required diagnostics. On the basis of the constructed wavelet bases, the results of the analyses of the signals registered in the different states of the winding short-circuits of the stator winding of an induction motor operating under the constant load, were presented. The choice of an advanced method of analysis was dictated by the occurrence of the periodic energy flow of signals between the adjacent frequency bands which made the Fourier analyses ineffective.