Preventive Maintenance of Motors and Automatic Classification of Defects using Artificial Intelligence

Abstract

Preventive maintenance of electromechanical systems avoids unexpected errors of motors, in which interior faults from wires, bearings, stator and rotors may induce rising temperature, mechanical deformation and structural vibration. Those coupled defects undermine efficiency and stability of rotational mechanism. Correct classification of those interior defects is one of key issues for preventive maintenance and accurate diagnosis is important work to precisely locating the poison of interior errors using sensors and algorithms. In this manner, this study presents a program for preventive maintenance of motors based on the application of support vector machine. Faults of unbalance, frictions and aging have been systematically embedded into bearings. Polyvinylidene fluoride and accelerometers have been used to retrieve time-displacement, time-acceleration, resonant frequencies. Those experimental results serve as training data for constructing hyper plane of the support vector machine. Decision based on the calibrated program automatically classifies defects of the motor system, and provides accurate result of 98 percent. The novelty of this study is not only the usages of thin film with piezoelectric sensors but also the practical applications of support vector machine with artificial intelligence.