Induction Motor Fault Diagnosis with Local Ternary Pattern and AI Approaches

Abstract

Owing to the induction machine's widespread use across most industries, engine failure will be quite expensive. To address this problem, numerous signal processing techniques have been adopted. This work proposed a novel technique called GLTP dependent on local ternary patterns for texture analysis to identify defects in induction motor with the grey level co-occurrence matrix (GLCM). This technique is compared with another original technique named GLKTP based on local ternary pattern using Kirsch operators used to show eight main directional changes in the image and combined with the GLCM matrix for texture analysis. A set of acoustic data is employed with different engine failures (bearing defects and broken bars). The multiclass MCSVM (Support Victor Machine One vs. All) and K-NN (K-nearest neighbourhood) and ANN artificial neural network classifiers are utilized for fault identification.