Publication

Abstract : A gearbox and its accessories are a significant component of an electromechanical system and any damage to the gear is likely to be fatal. Data set from a faulty and healthy gearbox are obtained and the gear faults namely gear tooth damage, bearing damage in gearbox and lubrication drain are investigated by analyzing the motor's current and voltage signature. A comparison between the conventional vibrational analysis and motor electrical signature analysis (MESA) is carried out using two specific machine-learning algorithms-Decision Tree Algorithm and Support Vector Machine (SVM) algorithm. It is observed that MESA is more accurate than the Vibrational analysis and SVM is a better classifier compared to decision tree. Both single and multiple faults are examined and SVM stands out for multiple fault detection in gear box.