Broken Bar Faults Detection Under Induction Motor Starting Conditions Using the Optimized Stockwell Transform and Adaptive Time–Frequency Filter
Mohamed Esam El-Dine Atta, Doaa Khalil Ibrahim, Mahmoud Gilany
Abstract
Most of the published research studies for detecting induction motor broken bar faults (BBFs) use a time-frequency (t - f) decomposition tool to characterize the fault-related components. However, the identification and the assessment of these components in (t - f) domain require skilled user or powerful pattern recognition technique. Moreover, a relatively long starting duration is necessary. This article introduces an automated scheme to detect BBFs and distinguish fault severity in induction motors under startup conditions regardless of the user experience and even under short starting duration and in a noisy environment. This scheme is based on the analysis of the starting current using optimized Stockwell transform (ST). An active-set algorithm is applied to maximize the energy concentration of the left-side harmonic (LSH) component. Then, an adaptive time-frequency filter is applied to extract the LSH component from the (t - f) domain, where the energy of the right part of LSH (RLSH) is utilized as an effective index for BBFs detection and for discriminating BBFs severity. Both real experimental data and simulation-based tests on 0.746- and 11-kW motors are used to extensively verify the performance of the proposed scheme. The achieved results have ensured that the proposed scheme can achieve a high accuracy with the minimum data and shortest acquisition time in comparison with some recent methods in the literature.