MEASUREMENT AND ANALYSIS OF ELECTRICAL MACHINE FAULTS

Abstract

In industry, continuous and trouble free operation of electric drives is an - important aspect. Faults and /or failures of electrical machinery not only cause economical losses but it may also create dangerous situations. Therefore, the measurement and analysis of electrical machine faults is necessary to diagnose the machine faults. Different methods for fault identification exist and used to detect the machine faults using machine parameters such as current, voltage, speed, temperature and vibrations. The primary goal of the present work is to develop measurement system and analyze the acquired signal. It is estimated that 40% of the motor faults are attributed to bearing faults. Detection of these defects is important, so vibration measurement and different signal processing techniques are employed for the detection and classification of bearing fault of induction motor. The vibration signal is analyzed in time, time- frequency domain and diagnostically important features are extracted. The extracted features are then used for the development of knowledge based model for classification purpose. To automate the diagnostic process, researchers have proposed intelligent diagnostic systems which uses artificial intelligence tools such as neural network, Support Vector Machines, Relevance Vector Machine, Fuzzy logic etc. In this work, for automatic detection of fault, a machine learning algorithm Support Vector Machine (SVM) is implemented. The SVM model is trained for different bearing faults taken from PBL-6308 database. Then, this designed SVM model was tested for different bearing faults taken from NBC-6308 database and CWRU, US database. SVM based classifier is built to minimize the structural misclassification risk, whereas conventional classification techniques often apply minimization of the empirical risk. It gives a better generalization capability and gives us global solution for a classification problem.