Using the goertzel algorithm over disjoint narrow frequency bands for fault diagnosis of induction motors

Abstract

© 2018 IEEE. Induction motors (IMs) are the driving force of modern industries. On-line condition monitoring of IMs tries to detect faults at early stage helping to adjust the maintenance schedule and thus, avoiding unexpected and costly breakdowns. Among the methods for condition monitoring, on-line motor current signature analysis is attracting a rising interest, because it is non-invasive and it can identify a wide variety of faults. Usually, motor current signature analysis (MCSA) approaches are Fourier based methods, which require long acquisition times (to achieve high spectral resolution) and high sampling frequencies (to reduce aliasing effects). However, these requirements generate a huge amount of data which can cause an overwhelming burden on embedded devices with very limited internal memory such as digital signal processors (DSPs) or field programmable gate arrays (FPGAs) among others. To address these issues, this paper introduces an approach which enables to carry-out fault diagnosis of IMs based on MCSA but using low cost computing devices, with limited memory and computing power. This approach proposes the use of disjoint narrow bandwidths and a modified sliding window fast Fourier transform (SFFT) based on the Goertzel algorithm. The modified SFFT proposed uses small data buffers to calculate the signal's spectrum. Besides, the use of disjoint narrow frequency bands highly reduces the number of bins being updated as each new data is sampled for fault diagnosis purposes. Based on this approach, an embedded device only needs to store and process a minimal set of bins, compared with other methods, for a desired resolution.