A Low-Complexity Algorithm for Diagnosis of Three-Phase Induction Motors

Three-phase motors are made up of different pieces and parts that, when worn, increase mechanical vibrations. These vibrations generate multiple frequencies and their respective harmonics, so detection and classification must be done in the frequency domain. The faults also depend on the shaft’s rotational frequency and the power supply frequency. State-of-the-art algorithms for motor failure detection and classification are complex since they involve processing in the time domain and the frequency domain. Given this, the present work proposes a low complexity algorithm based on wireless monitoring of certain frequency components’ amplitude to define the degree of severity of vibrations in a three-phase motor. The algorithm consists of filtering the signal through a bank of low-pass filters, obtaining frequency spectra through the FFT, averaging the modulus spectra to reduce noise and distortion, and evaluating the amplitudes of the components of the frequency of failure in order to detect and identify the degree of severity based on reference levels of velocity. The algorithm offers a quantitative-qualitative diagnosis because based on an amplitude (quantitative), we designate a level of severity (qualitative). Now the technician offers a qualitative diagnosis - qualitative. In this case, the technician’s verbal coincidences determine the level of severity according to ISO 20816-1: 2016. Then the coincidences at the level of severity (qualitative) establish the certainty of the algorithm developed concerning the technician’s diagnosis. The diagnosis coincidence establishes a result with a high level of reliability based on the similarity between the technician’s diagnosis and the algorithm.