Model-Based Diagnostics and Fault Assessment of Induction Motors with Incipient Faults Mohsen Nakhaeinejad1, Jaewon Choi2, Michael D. Bryant3 Department of Mechanical Engineering The University of Texas at Austin, Texas 78712 1 2 3 ABSTRACT Model-based diagnostics is relatively new for machine condition monitoring. Unlike signal-based approaches, health metrics are evaluated via physics-based models and sensor measurements. In this article, a framework of model-based fault diagnosis and severity assessment for induction motors is presented. A conventional symmetric induction motor model is suitable for simulation but cannot capture faulty behavior accurately. Induction motor is modeled in detail using vector bond graph technique. Staged faults are induced in simulation to emulate broken rotor bar, stator winding fault and bearing faults. Extended Kalman Filter (EKF) technique estimates and tunes parameters of the model to detect faults. Changes of the model parameters pinpoint degradation and faults of the actual system. Fault severity assessment is performed using Channel capacity technique, a novel heath metric based on Shannon’s information theory. Results suggest that stator winding faults and broken rotor bars can be detected with model-based diagnostics. Keywords: Model-Based Diagnostics, Condition Monitoring, Induction Motor, Extended Kalman Filter, Health Assessment, Fault Isolation 1 Model Based Diagnostics Machines break down due to wear, crack, lubricant issues, external impacts, faulty electronics, and so on. The economic losses often exceed the repair expenses itself [1]. Machine maintenance requires expertise, effort, time and most of all perseverance. Industry requires diagnostics packages, which might be constructed based on advanced diagnostics algorithm. But, they should be designed in such a way to be used by technicians. Faults and degradation in machines are related to the change of states, parameters, or process. A model-based diagnostic system consisting of physics-based model, parameter tuning module, and decision making unit tracks changes of states and parameters to detect degradation and faults. Model Based Diagnostics (MBD) is based on fundamentals as tools for fault detection and isolation (FDI) [2]. Different from signal based approaches [3], MBD relies on constitutive laws of physics and T. Proulx (ed.), Rotating Machinery, Structural Health Monitoring, Shock and Vibration, Volume 5, Conference Proceedings of the Society for Experimental Mechanics Series 8, DOI 10.1007/978-1-4419-9428-8_37, © The Society for Experimental Mechanics, Inc. 2011 439 mohsenn@mail.utexas.edu, jaewon.choi.eng@gmail.com, mbryant@mail.utexas.edu
RkJQdWJsaXNoZXIy MTMzNzEzMQ==