Search

This paper proposes a novel brushless excitation topology for a three-phase synchronous machine based on a customary current-controlled voltage source inverter (VSI). The inverter employs a simple hysteresis-controller-based current control scheme that enables it to inject a three-phase armature current to the stator winding which contains a dc offset. This dc offset generates an additional air gap magneto-motive force (MMF). On the rotor side, an additional harmonic winding is mounted to harness the harmonic power from the air gap flux. Since a third harmonic flux is generated in this type of topology, the machine structure is also modified to accommodate the third harmonic rotor winding to have a voltage induced as the rotor rotates at synchronous speed. Specifically, four-pole armature and field winding patterns are used, whereas the harmonic winding is configured for a twelve-pole pattern. A diode rectifier is also mounted on the rotor between the harmonic and field windings. Therefore, the generated voltage on the harmonic winding feeds the current to the field winding for excitation. A 2D-finite element analysis (FEA) in JMAG-Designer was carried out for performance evaluation and verification of the topology. The simulation results are consistent with the proposed theory. The topology could reduce the cost and stator winding volume compared to a conventional brushless machine, with good potential for various applications. [ABSTRACT FROM AUTHOR]

Convolutional Neural Networks (CNNs) and Deep Learning (DL) revolutionized numerous research fields including robotics, natural language processing, self-driving cars, healthcare, and others. However, DL is still relatively under-researched in physics and engineering. Recent works on DL-assisted analysis showed enormous potential of CNN applications in electrical engineering. This paper explores the possibility of developing an end-to-end DL analysis method to match or even surpass conventional analysis techniques such as finite element analysis (FEA) based on the ability of CNNs to predict the performance characteristics of electric machines. The required depth in CNN architecture is studied by comparing a simplistic CNN with three ResNet architectures. Studied CNNs show over 90% accuracy for an analysis conducted under a minute, whereas a FEA of comparable accuracy required 200 h. It is also shown that training CNNs to predict multidimensional outputs can improve CNN performance. Multidimensional output prediction with data-driven methods is further discussed in context of multiphysics analysis showing potential for developing analysis methods that might surpass FEA capabilities. [ABSTRACT FROM AUTHOR]