Bond Graph Modelling for Condition Monitoring of Induction Motors

The complication of existing electromechanical systems and the increased demands on their operational performances of efficiency and reliability motivate the need for monitoring and fault diagnosis of these systems. Motor Current Analysis (MCA) is a cost-effective technique for the detection of motor faults. To our knowledge, MCA has not been used with bond graph (BG) modeling for developing accurate diagnostic information. In this paper, a BG model is developed for fault detection of AC Induction Motors (ACIM) based on motor current analysis. BG is a single language for unified domains, which allows the dynamics of electrical and mechanical effects to be modeled directly. In the proposed model the physical components of the electro-mechanical system are constructed by including three different levels of modeling, conceptual behavior, cause and effect relations, and numerical model. This BG model was examined based on the behavior of the ACIM and confirmed the high efficiency of BG based approach in achieving diagnostics of different fault cases. In particular, the focus is on the impact of both the broken rotor bars (BRB) and stator short circuit (SSC) that commonly occur in ACIM. The simulation results indicate that the proposed BG approach is an effective method for extracting diagnostic information based on current analysis. The relationship between the sideband components and the system behavior can be used as an indicator to distinguish between healthy condition, BRB and SSC. The results were evaluated using experiments data. Faults in ACIM are investigated actively.